HIGH SPEED AND HIGH EFFICIENCY INFRARED · PDF fileHIGH SPEED AND HIGH EFFICIENCY INFRARED...

HIGH SPEED AND HIGH EFFICIENCY INFRARED

PHOTODETECTORS

A DISSERTATION

SUBMITTED TO THE DEPARTMENT OF PHYSICS

AND THE INSTITUTE OF ENGINEERING AND SCIENCES

OF BİLKENT UNIVERSITY

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

By

İBRAHİM KİMUKİN

DECEMBER 2004

I certify that I have read this thesis and that in my opinion it is

fully adequate, in scope and in quality, as a dissertation for the

degree of Doctor of Philosophy.

________________________________________

Prof. Ekmel Özbay (Supervisor)




________________________________________

Asst. Prof. Oğuz Gülseren




________________________________________

Assoc. Prof. Ahmet Oral




________________________________________

Asst. Prof. Özgür Aktaş




________________________________________

Prof. Tayfun Akın

Approved for the Institute of Engineering and Sciences:

________________________________________

Prof. Mehmet Baray,

Director of Institute of Engineering and Sciences

Abstract

HIGH SPEED AND HIGH EFFICIENCY INFRARED

PHOTODETECTORS

İbrahim Kimukin

Ph. D. in Physics

Supervisor: Prof. Ekmel Özbay

December 2004

The increasing demand for telecommunication systems resulted in production

of high performance components. Photodetectors are essential components of

optoelectronic integrated circuits and fiber optic communication systems. We

successfully used resonant cavity enhancement technique to improve InGaAs

based p-i-n photodetectors. The detectors had 66% peak quantum efficiency at

1572 nm which showed 3 fold increases with respect to similar photodetector

without resonant cavity. The detectors had 28 GHz 3-dB bandwidth at the same

time. The bandwidth efficiency product for these detectors was 18.5 GHz, which is

one of the best results for InGaAs based vertical photodetector.

The interest in high speed photodetectors is not limited to fiber optic networks.

In the recent years, data communication through the air has become popular due to

i

ease of installation and flexibility of these systems. Although the current systems

still operate at 840 nm or 1550 nm wavelengths, the advantage of mid-infrared

wavelengths will result in the production of high speed lasers and photodetectors.

InSb based p-i-n type photodetectors were fabricated and tested for the operation

in the mid-infrared (3 to 5 µm) wavelength range. The epitaxial layers were grown

on semi-insulating GaAs substrate by molecular beam epitaxy method. The

detectors had low dark noise and high differential resistance around zero bias.

Also the responsivity measurements showed 49% quantum efficiency. The

detectivity was measured as 7.98×109 cm Hz1/2/W for 60 µm diameter detectors.

Finally the high speed measurements showed 8.5 and 6.0 GHz bandwidth for 30

µm and 60 µm diameter detectors, respectively.

Keywords: Photodetector, high speed, high efficiency, resonant cavity

enhancement, infrared wavelengths.

ii

Özet

YÜKSEK HIZLI VE YÜKSEK VERİMLİ KIZILÖTESİ

FOTODEDEKTÖRLER

İbrahim Kimukin

Fizik Doktora

Tez Yöneticisi: Prof. Ekmel Özbay

Aralık 2004

Telekomnikasyon sistemlerine ilginin artması yüksek performanslı elemanların

üretimiyle sonuçlanmıştır. Fotodedektörler optoelektronik entegre devrelerin ve

fiber iletiim sistemlerinin önemli bir parçasıdır. Resonant çınlaç arttırımı

tekniğinini InGaAs temelli p-i-n tipi fotodedektörleri geliştirmek için kullandık.

Dedektörlerin 1572 nm’deki %66 tepe kuantum verimi rezonatörsüz benzer bir

dedektöre göre 3 kat fazladır. Dedektörler aynı zamanda 28 GHz 3-dB bantralığına

iii

sahiptir, ki bu değerler InGaAs temelli düşey fotodedektörler için en iyi

sonuçlardan biridir.

Yüksek hızlı fotodedektörlere ilgi sadece fiber optic ağlar ile sınırlı değildir.

Yakın zamanda havadan data iletimi kolay kurulum ve değişken yapılarından

dolayı populer hale gelmiştir. Hernekadar şu anda kullanılan sistemler 840 nm ve

1550 nm de çalışsa da, orta-kızılötesi dalagaboylarının avantajı yüksek hızlı lazer

ve fotodedektörlerin üretimine sebep olacaktır. InSb temelli p-i-n tipi dedektörler

orta-kızılötesi dalagaboylarında (3 ten 5 µm’ye) kullanılmak üzere üretilip

testedildi. Dedektör katmanları yarı-yalıtkan GaAs altaşın üzerine moleküler ışın

demeti metodu ile büyütüldü. Dedektörler sıfır volt civarında düşük karanlık akım

ve yüksek dirence sahiptiler. 60 µm çaplı dedektörlerin hassasiyeti 7.98×109 cm

Hz1/2/W olarak ölçüldü. Son olarak yüksek hız ölçümlerine göre 30 µm ve 60 µm

çaplı dedektörlerin hızları sırasıyla 8.5 GHz ve 6.0 GHz’tir.

Anahtar Kelimeler: Fotodedektör, yüksek hız, yüksek verim, resonant çınlaç

arttırımı, kızılötesi dalgaboyları.

iv

Acknowledgements

It is my pleasure to express my deepest gratitude to my supervisor Prof. Ekmel

Özbay for his invaluable guidance, motivation, encouragement, confidence,

understanding, and endless support. My graduate study under his supervision was

my lifetime experience. It was an honor to work with him, and I learned a lot from

his superior academic personality.

I would like to thank Prof. Orhan Aytür and Dr. Tolga Kartaloğlu for their help

in high speed measurements in the infrared wavelengths. I would like to thank Prof.

Cengiz Beşikçi, Selçuk Özer, and Orkun Cellek for their help in responsivity

measurements of InSb photodetectors.

I also thank the members of the thesis committee, Asst. Prof. Oğuz Gülseren,

Assoc. Prof. Ahmet Oral, Asst. Prof. Özgür Aktaş, and Prof. Tayfun Akın for their

useful comments and suggestions.

I would like to thank all the former and present members of Advanced

Research Laboratory for their continuous support. I want to especially thank the

group members of the detector research team: Necmi Bıyıklı, Bayram Bütün,

v

Turgut Tut. It was a pleasure to work with these hard-working guys in the same

group. Other special thanks go to Murat Güre and Ergün Kahraman for their

technical help and keeping our laboratory in good condition.

Finally I would express my endless thank to my family for their understanding

and continuous moral support. Very special thanks belong to my wife, Sevcan, for

her endless moral support, understanding, and love.

vi

Contents

Abstract

Özet

Acknowledgements

Contents

List of Figures

List of Tables

1 Introduction

2 Theoretical Background

2.1 Photodetectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.1.1 Photodetector Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.1.2 Photodetector Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

i

iii

v

vi

xi

xvii

1

5

5

6

10

vii

2.1.3 Photodetector Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.2 Resonant Cavity Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.2.1 RCE Formulation and Optimization . . . . . . . . . . . . . . . . . . . . . .

2.2.2 Standing Wave Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.3 Optical Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.3.1 Transfer Matrix Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.3.2 Distributed Bragg Reflectors . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 Fabrication and Characterization

3.1 Basic Fabrication Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.1 Sample Cleaving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.2 Sample Cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.1 Photolithography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.4 Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.5 Etch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.6 Metallization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.7 Lift-off . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.8 Thermal Annealing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.9 Dielectric Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.2 Measurement Setups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.2.1 Reflection and Transmission Measurements . . . . . . . . . . . . . . . .

3.2.2 Current Voltage Measurements . . . . . . . . . . . . . . . . . . . . . . . . . .

3.2.3 Quantum Efficiency Measurements . . . . . . . . . . . . . . . . . . . . . .

3.2.4 High Speed Measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12

15

15

18

19

19

23

26

26

26

27

28

31

32

35

36

37

38

41

41

42

43

45

viii

4 InGaAs p-i-n Photodetector

4.1 Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.2 Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.2.1 Large Area Photodetector Fabrication . . . . . . . . . . . . . . . . . . . .

4.2.2 Small Area Photodetector Fabrication . . . . . . . . . . . . . . . . . . . .

4.3 Measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.3.1 Current-Voltage Characteristics . . . . . . . . . . . . . . . . . . . . . . . . .

4.3.2 Responsivity Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.3.3 High-Speed Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 InSb p-i-n Photodetector

5.1 Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.2 Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.3 Measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.3.1 Current-Voltage Characteristics . . . . . . . . . . . . . . . . . . . . . . . . .

5.3.2 Responsivity Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.3.3 High Speed Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6 Conclusion and Suggestions for Further Research

Bibliography

A Frequency Response Calculation for a Photoetector

B TMM Based Detector Design

47

49

58

60

65

68

70

75

76

79

81

84

85

91

91

95

99

102

106

109

121

122

ix

C List of Publications

126

x

List of Figures

2.1

2.2

2.3

2.4

2.5

2.6

2.7

2.8

2.9

2.10

2.11

2.12

2.13

3.1

3.2

3.3

Energy diagram of a p-i-n photodetector . . . . . . . . . . . . . . . . . . . . . . . .

Energy diagram of a Schottky photodetector . . . . . . . . . . . . . . . . . . . . .

Cross-section of vertical illuminated photodetectors . . . . . . . . . . . . . . .

Cross-section of edge illuminated photodetectors . . . . . . . . . . . . . . . . .

Output current versus time for a constant illumination across the

depletion region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Small signal equivalent circuit for photodetector . . . . . . . . . . . . . . . . . .

Generalized structure of a resonant cavity enhanced photodetector . . . .

Electric field amplitudes at the interface . . . . . . . . . . . . . . . . . . . . . . . .

The propagation of wave inside the same medium . . . . . . . . . . . . . . . . .

Electric field values before and after the structure . . . . . . . . . . . . . . . . .

Reflection spectrum and phase difference of InP/Air DBR with 3 pairs

Reflection spectrum and phase difference of InP/InAlGaAs DBR with

40 pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Transmission spectrum of Si/SiO2 notch filter centered to 1550 nm . . .

Photographs of some samples with fabricated photodetectors on them

Photographs of the masks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Plots from the mask file (a) a photodiode with 100 µm diameter active

7

8

11

11

12

13

16

20

21

22

24

24

25

27

28

xi

3.4

3.5

3.6

3.7

3.8

3.9

3.10

3.11

3.12

area, (b) development marks used in the p+ ohmic step, (c) p+ ohmic

transmission lines and the development marks, and (d) alignment

marks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Plot of the mask containing large area photodetectors. All the

fabrication steps are shown aligned to each other. The mask contains

around 115 photodetectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Image-reversal photolithography (after reference [30]) . . . . . . . . . . . . .

SEM picture wet etch profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) The current flow between the contacts. (b) The schematics of the

model used for the ohmic contact characterization. (c) The p+ ohmic

transmission line (center) is used for ohmic quality measurements.

Also the alignment marks (up) and the development marks (bottom)

can be seen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Pictures of equipments: (a) mask aligner, (b) surface profilometer, (c)

thermal evaporator and RF sputter, (d) PECVD, (e) RIE, (f) RTA . . . .

(a) The reflectivity measurement setup. The light source (blue) the

fiber probe and the optical spectrum analyzer can be seen. (b) The

reflectivity of the InGaAs based photodetector wafer . . . . . . . . . . . . . .

(a) The probe station. With the help of objectives, we can make

contact to the pads of the photodetectors. (b) The sample on the probe

station, where two DC probes touch the pads, while the active are is

illuminated with the fiber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

HP4142B and the computer control . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Quantum efficiency measurement setup with the monochromator . . . .

29

30

31

33

38

40

42

43

43

44

xii

3.13

3.14

4.1

4.2

4.3

4.4

4.5

4.6

4.7

4.8

4.9

4.10

Quantum efficiency measurement setup with the tunable laser . . . . . . .

High Speed measurement setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Lattice parameter vs. energy gap of III-V compounds used with InP

and GaAs based devices (after reference [61]) . . . . . . . . . . . . . . . . . . . .

Charge drift velocity as a function of electric filed strength for various

semiconductors (after reference [61]) . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) Lattice match conditions of quaternary compounds to InP (after

reference [68]). (b) Energy gap of III-V compounds (after reference

[69]) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Refractive index of (In0.52Al0.48As)χ(In0.53Ga0.47As)1-χ as a function of

wavelength for different values of χ (after reference [70]) . . . . . . . . . . .

Spectral reflectivity of the bottom DBR calculated using TMM . . . . . .

The measurement (solid line) and calculation (dotted line) results for a

region on the wafer where the shift was %4 . . . . . . . . . . . . . . . . . . . . . .

The n+ ohmic contacts before (a) and after (b) the RTA . . . . . . . . . . . .

The p+ ohmic contacts before the metal deposition (a) and after the

RTA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) The color change on the surface is due to the variation of the etch

depth (b) the mesa contour can be seen as a thin black line below the p

contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) The surface was covered with silicon nitride, and the photograph

shows the surface after photolithography, (b) the surface of the sample

after the HF etch and the cleaning. Some area on top of the contacts

was open, and the sidewalls were covered . . . . . . . . . . . . . . . . . . . . . . .

45

46

48

52

53

55

56

59

61

62

62

63

xiii

4.11

4.12

4.13

4.14

4.15

4.16

4.17

4.18

4.19

4.20

4.21

(a, b) The photoresist before the metallization, (c, d) interconnect

metals after the cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

SEM images of large area photodetectors . . . . . . . . . . . . . . . . . . . . . . .

Cross-section of the sample (a) after airbridge metallization, and (b)

after the liftoff of the metal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

SEM picture of airbridges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3D illustration of a fabricated photodetector . . . . . . . . . . . . . . . . . . . . .

Pictures of small area photodetectors obtained with optical microscope

and SEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) Current voltage characteristics of a 30 µm diameter photodetector,

where the breakdown can be seen around -14 V bias. (b) The

logarithmic plot of I-V characteristics of 30, 60, 100, 150, and 200 µm

diameter photodetector (from bottom to top respectively) . . . . . . . . . . .

(a) Dark current of photodetectors at 1 V reverse bias as a function of

detector area. (b) Dark current density of photodetectors at 1 V reverse

bias as a function of detector area. The linear decrease can be seen for

detectors having 60 µm diameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) Calculated differential resistance of the photodetectors with 30, 60,

100, 150, and 200 µm diameter (from top to bottom respectively). (b)

The maximum differential resistance of photodetectors as a function of

detector area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) Calculated differential resistance of the photodetectors at zero bias.

(b) R0A product as a function of detector area . . . . . . . . . . . . . . . . . . . .

(a) The spectral quantum efficiency measurement for successive recess

64

65

67

68

68

69

71

72

73

74

xiv

4.22

4.23

5.1

5.2

5.3

5.4

5.5

5.6

5.7

5.8

5.9

5.10

etches. (b) The measurement and calculation of the spectral quantum

efficiency with the peak at 1572 nm . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) The photocurrent vs optical power measurement of the

photodetectors under various bias voltages. (b) Log-Log plot of the

previous data for 0 and 4 V bias voltages . . . . . . . . . . . . . . . . . . . . . . . .

(a) The temporal response of 5×5 µm2 area photodetector under 7 V

reverse bias. (b) The calculated frequency response of the

photodetector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Spectral transmission of the atmosphere at the sea level . . . . . . . . . . . .

Lattice constant and energy gap of semiconductors . . . . . . . . . . . . . . .

Details from InSb wet etch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Modified mesa isolation step used during the fabrication of InSb

photodetectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) Cross-section of a fabricated photodetector (b) Photograph of

photodetector with 150 µm diameter . . . . . . . . . . . . . . . . . . . . . . . . . . .

Current-voltage characteristics and differential resistance of 30 µm

diameter photodetector at 300 K (dotted line) and 77 K (solid line) . . .

Measured current-voltage characteristics of 60, 100, and 150 µm

diameter photodetectors at 77 K . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) Dark current at -0.5 V as a function of area, while (b) shows the

dark current density (J0) as a function of area . . . . . . . . . . . . . . . . . . . .

Calculated differential resistance (Rd) as a function of bias voltage . . .

(a) The zero bias differential resistance (R0) as a function of detector

area. (b) Resistance-area product (R0A) as a function of detector area .

77

78

80

81

85

87

88

90

92

92

93

94

95

xv

5.11

5.12

5.13

5.14

5.15

5.16

5.17

5.18

Responsivity of the photodetectors at 1550 nm as a function of the

reverse bias. The inset shows the spectral responsivity measurement

under different reverse bias voltages ranging from 0.2 to 0.6 V . . . . . .

Spectral simulation results for optical reflection and absorption in the

p+ and n- InSb layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Picture of the setup from Electrical Engineering of METU . . . . . . . . . .

Results of the spectral responsivity measurements are shown for 80

µm (solid line) diameter and 60 µm (dotted line) diameter

photodetector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(a) The schematic diagram of the optical parametric oscillator, (b)

pictures of the experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Temporal response of a 30 µm diameter detector under (a) 0.5V, (b)

1.0V, and (c) 2.5V bias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Temporal response of a 60 µm diameter detector under (a) 0.5V, (b)

1.0V, and (c) 2.5V bias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Fast Fourier Transform of the temporal responses of the

photodetectors. Results for (a) 60 µm diameter and (b) 30 µm diameter

photodetectors as a function of bias are shown. (c) 3-dB bandwidth of

the 30 and 60 µm diameter photodetectors as a function of applied bias

96

97

98

99

100

103

104

105

xvi

List of Tables

3.1

3.2

3.3

4.1

4.2

4.3

4.4

4.5

5.1

5.2

5.3

Etch rates of some recipes used in our processes . . . . . . . . . . . . . . . . . .

Etch rates of some recipes used in RIE etch . . . . . . . . . . . . . . . . . . . . . .

Growth conditions and properties of dielectric films . . . . . . . . . . . . . . .

Comparison of this work with the previous research . . . . . . . . . . . . . . .

Optical properties of various III-V semiconductor compounds . . . . . . .

Band discontinuities of some heterostructure systems . . . . . . . . . . . . . .

Epitaxial structure of the InGaAs based pin detector wafer . . . . . . . . . .

Etch rates of some common etchants. The rates are indicated as

nm/sec. Stop means no etch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Gases and their absorption properties . . . . . . . . . . . . . . . . . . . . . . . . . . .

Properties of some low energy gap semiconductors . . . . . . . . . . . . . . . .

Epitaxial layers of InSb p-i-n photodetector . . . . . . . . . . . . . . . . . . . . . .

33

34

41

50

51

52

57

61

82

83

85

xvii

Chapter 1

Introduction

The invention of the telephone led to an enormous world wide

telecommunication network. Especially during and after the Second World

War, the demand for the calling services resulted to both cheaper and faster

services. The invention of the first solid-state transistor in 1947, and then the

integrated circuit opened the age of computing and communication [1]. In

1960's researchers developed the first laser [2].

The development of the first commercially feasible optical fiber in

1970's made the fiber optic communication a promising candidate for

telecommunication [3]. In the early 1980's satellites capable of carrying nearly

100,000 simultaneous calls are used for telephone calls. Demand for the faster,

cheaper, and less noisy communication made the first transatlantic undersea

fiber-optic telephone cable possible that replaced the copper one that had been

installed in 1956. That was the first fiber-optic revolution in

telecommunication.

The second revolution came with the introduction of wavelength

division multiplexing (WDM). Telephone companies has laid cables containing

24 to 36 fibers with 2.5 Gbit/s rate, many had been reserved as “dark fiber”.

But the tremendous traffic has crowded these cables that once seemed so

1

CHAPTER 1. INTRODUCTION 2

voluminous. This demand was due to the growth of the Internet, and the

demand for more information transform for entertainment and communication.

In the mid 1990's, companies began using systems capable of transmitting at

four wavelengths, and soon this number increased to eight. Nowadays, systems

with 40 channels of 10 Gb/s, or 80 channels of 2.5 Gb/s are available to be

used with dispersion managed fibers [4].

This high demand for better fiber optics communication systems led

scientist and engineers to work on components like lasers, modulators,

photodetectors, optical amplifiers, and optical fibers. The optical fiber offers an

operation bandwidth up to tens of THz. The research effort in optoelectronics

is devoted to fully exploit the fiber bandwidth. This can be possible with high

performance components.

Due to the properties of the commercial silica based fibers,

optoelectronic research focused at three wavelengths, where the minimum

attenuation occurs [5]. First one is located at 850 nm, which is called the first

optical window. GaAs based detectors are usually used for the detection. Local

area networks use this window, because the high loss in this wavelength

prohibits the communication for long distances. These systems need only

multimode fiber and transmitter, hence they are cheap. On the other hand, the

second and the third windows are located at 1310 and 1550 nm respectively.

The loss in these wavelengths is much less than the first optical window. Due

to low loss, these wavelengths are used for long distance fiber optic

communication. These systems require high speed modulator, detectors, and

repeaters. Semiconductor based photodiodes demonstrate excellent features to

fulfill the requirements of high speed optoelectronic receivers. GaAs and

InGaAs are the most studied materials for high speed photodetection.

Photodetectors have been demonstrated bandwidth capabilities as high as 200


GHz [6-9]. However, the efficiencies of these detectors have been less than

10%, due to thin absorption layer needed for short transit time.

Resonant cavity enhanced (RCE) photodetectors offer the possibility of

overcoming this limitation of bandwidth-efficiency product of conventional

photodetectors [10]. The RCE photodetectors are based on the enhancement of

the optical field inside a Fabry-Perot cavity. This enhancement allows the

usage of thinner active layers, which minimizes the transit time without

sacrificing the quantum efficiency. This thesis reports the development of high

efficiency photodetectors with the use of RCE effect in variety of our detectors.

Chapter 2 reviews the theory of photodetectors. It also examines the

transport of carriers, and the high speed design. The chapter summarizes the

theory of resonant cavity enhancement (RCE) and presents the simulation

results of the RCE photodetectors. Finally it includes the techniques used for

the optical design of multilayer photodetectors.

Chapter 3 gives the details of the fabrication steps. Details of the

fabrication steps are also given. This chapter also includes the measurement

setups and procedures used during the characterization.

Chapter 4 explains the design, fabrication and characterization of

InGaAs based pin type photodetector. This detector includes RCE design for

the operation around 1550 nm. The fabrication procedure is given with details.

The results of the current-voltage, responsivity, and high-speed measurements

are given at the characterization part.

Chapter 5 is devoted to InSb based photodetectors. Design, fabrication

and measurement details are presented. This detector was designed for the

operation in the 3–5 µm wavelength range, so that can be used for detection in

the mid-infrared region. The design and fabrication were done at Physics

Department of Bilkent University. Some of the measurements were done in


collaboration with the Electrical and Electronics Engineering Department of

Bilkent University and Middle East Technical University.

Chapter 2

Theoretical Background

This chapter briefly explains the photodetector operation and the resonant

cavity enhancement (RCE) technique that is used to achieve higher

responsivity. Also design basics are presented. Section 2.1 explains the

operation of a semiconductor junction photodetector. Basic formulation of the

RCE technique is presented in Section 2.2. Finally, Section 2.3 explains the

design basics and calculation methods of optical properties of multi-layered

structures.

2.1 Photodetectors Photodetectors can be classified into two categories: thermal detectors and

quantum detectors. Thermal detectors sense the radiation by its heating effect.

Thermal detectors have the advantage of wide spectral range and operation at

the room temperature, but they are limited as far as speed and sensitivity are

concerned. Bolometers, thermistors, pyroelectric detectors are widely used

thermal detectors.

Operation of quantum detectors depend on the discrete nature of

photons which can transfer energy of individual particles (electron and hole)

giving rise to a photocurrent. These devices are characterized by very high

sensitivity, high speed but limited spectral response. Photoemissive,

5

CHAPTER 2. THEORETICAL BACKGROUND 6

photovoltaic, and photoconductive detectors are classified in this category. The

conductivity change in the photoconductive detectors is proportional to the

intensity of the radiation falling on the semiconductor. Rectifying junctions are

the basis of the photovoltaic detectors. Under the illumination, the excess

carriers created within the semiconductor generate a proportional output

current. In our work, we mostly design and fabricate photovoltaic detectors.

2.1.1 Photodetector Basics A photovoltaic photodetector contains a depleted region with high electric field.

When photons are absorbed, charged carriers are generated inside the depletion

region. The photogenerated electron holes move in opposite directions until

they reach a highly doped contact layer or recombine. While the charges move

in the depletion region, a potential difference is created across it. This causes a

current flow through the circuit that the detector is connected. The depletion

region can be formed in different ways. Most popular are the p-n and metal-

semiconductor junctions.

The most popular photodetector type with a p-n junction is p-i-n

photodetector. In this structure, a lightly doped layer is placed between two

highly doped layers, where ohmic contacts are made. The lightly doped layer is

usually chosen to be n-type, hence a depletion region starting from the p-i

junction is formed. The width of the depletion region can be controlled with the

doping and applied voltage.

The highly doped layers are grown with a semiconductor having higher

energy band than the intrinsic region. In this way, all the incoming light is

absorbed in i-layer where the electric field intensity is high. As a result, the

diffusion of photogenerated carriers from the highly doped regions is not

present, and the speed of the photodetector is only limited by the transit time of


carriers and the junction capacitance. The only difficulty in such hetero

junctions is the band discontinuities. If the discontinuity is high, charge

trapping at the interfaces degrades the performance of the photodetector.

Figure 2.1 shows an energy band diagram of a homojunction p-i-n

photodetector.

Ef

Ec

Ev

p+ region n- region n+ region

Wdepletion

region

Figure 2.1: Energy diagram of a p-i-n photodetector.

When we investigate the current-voltage characteristics of the p-i-n

photodetector, we see that the current is due to the sum of the diffusion and the

generation-recombination currents inside the intrinsic region. Then the current

density can be written as: 2 2

exp 1 exp 12 2

p i n i iA A

p D n A

qD n qD n qWnqV qVJL N L N kT kTτ

⎛ ⎞ ⎡ ⎤ ⎡⎛ ⎞ ⎛ ⎞= + − +⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎤

−⎢ ⎥ ⎢⎜ ⎟ ⎝ ⎠ ⎝ ⎠ ⎥⎣ ⎦ ⎣⎝ ⎠ ⎦

(2.1)

Here W is the width of the depletion region, Dn and Dp are the diffusion

coefficient of minority carriers, Ln and Lp are the diffusion lengths, Nd and Na

are the doping concentration of the n+ and p+ regions. ni is the intrinsic carrier


concentration, τ is the carrier lifetime, and q is the electron charge. Usually NA

and ND are very high, and the diffusion current is negligible. Forward current

has an exponential dependence on the applied voltage. The reverse current

(dark current if there is no illumination) is constant and given by:

τ2i

RGqWnJ =−

(2.2)

Ef

Ec

Ev

metal semiconductor

Wdepletion

region

φB

Figure 2.2: Energy diagram of a Schottky photodetector.

On the other hand, a Schottky type photodetector consists of a metal-

semiconductor junction and the depletion region. The theory of rectification in

metal-semiconductor junction was developed in 1930’s by W. Schottky who

attributed rectification to a space-charge layer in the semiconductor [11].

Figure 2.2 shows the energy band diagram after the contact is made and the

equilibrium has been established. When the two substances are brought into

intimate contact, electrons from the conduction band of the semiconductor,

which have higher energy than the electrons in the metal, flow into the metal.


This process continues until the Fermi level on both sides is brought into

coincidence. As the electrons move from semiconductor to metal, the free

electron concentration in the semiconductor region near the boundary

decreases. The length of this region is given by [12]:

)(2 VVqN

W bid

−=ε

(2.3)

where, Nd is the ionized donor density, ε is the dielectric constant of the

semiconductor, Vbi is the built-in potential, V is the applied voltage. As the

separation between the conduction band edge Ec and the Fermi level Ef

increases with decreasing electron concentration and in thermal equilibrium Ef

remains constant, the conduction band edge bends up. The conduction band

electrons which cross over the metal leave a positive charge of ionized donor

atoms behind, so the semiconductor region near the metal gets depleted of

mobile electrons. Consequently an electric field is established from the

semiconductor to the metal. The build-in potential due to that electric field is

given by the difference of the work function of metal (φm) and semiconductor

(φs):

smbiqV φφ −= (2.4)

The current transport in the metal-semiconductor contacts is mainly due

to majority carriers, in contrast to p-n junctions. There are four different

mechanisms by which the carrier transport can occur: (1) thermionic emission

over the barrier, (2) tunneling through the barrier, (3) carrier recombination (or

generation) in the depletion region, and (4) carrier recombination in the neutral

region of the semiconductor. Usually the first process is the dominant

mechanism in Schottky barrier junctions in Si and GaAs and leads to the ideal

diode characteristics.


The thermionic emission theory is derived by Bethe [13] for high-

mobility semiconductors, and the diffusion theory is derived by Schottky [11]

for low-mobility semiconductors. A synthesis of the thermionic emission and

diffusion approaches has been proposed by Crowell and Sze [14]. The

complete expression of the J-V characteristics is given by:

⎥⎦⎤

⎢⎣⎡ −⎟

⎠⎞

⎜⎝⎛= 1exp

nkTqVJJ A

s

(2.5)

⎟⎟⎠

⎞⎜⎜⎝

⎛−= ∗∗

kTqTAJ B

sφexp2

(2.6)

where A** is the effective Richardson constant.

Some metal-semiconductor-metal (MSM) photodetectors have Schottky

contacts while some of them have ohmic contacts between the metal and

semiconductor.

2.1.2 Photodetector Structures Conventionally, the photodetectors are illuminated vertically, i.e., normal to the

epitaxial layers. Usually incoming light first travels through the detector layers,

and the optical power not absorbed pass through the substrate, as shown in

Figure 2.3(a). In some cases light is launched from the bottom, initially through

the substrate (usually thinned for lower absorption) and finally through the

detector layers (as shown on Figure 2.3(b)). This type of illumination is very

easy to couple all the incident power into the active area of the photodetector.

In some cases, the top contact (either ohmic or Schottky) is used as the mirror.

In this case optical power passes through the active layer twice; hence the

quantum efficiency is increased.


p+

i

n+

substrate

d

(a)

p+

i

n+

substrate

d

(b) Figure 2.3: Crossection of vertical illuminated photodetectors.

The other possible illumination is from the edge of the wafer as shown

on Figure 2.4. The width and the thickness of the waveguide must be small to

operate the waveguide in the singe mode. In this configuration, the coupling of

the incident power into the waveguide is the main problem. The coupled light

travels along the waveguide, generating electron hole pairs. If the length is high

enough, all the power can be absorbed in the waveguide. The loss due to the

absorption in the contact metals must be minimized.

p+

i

n+

substrate

d

Figure 2.4: Crossection of edge illuminated photodetectors.


2.1.3 Photodetector Operation We observe the transit of the photogenerated carriers inside the

depletion region. We start with an optical excitation at some specific point.

Modeling the depletion region as a parallel plate capacitor, we obtain the

current flowing out of the capacitor as:

⎪⎩

⎪⎨

⎧

<<=

<<+==

ehe

hhe

outtttv

dqI

ttvvdqI

tI,

0,)()(

2

1 (2.7)

Here d is the width of the depletion region; ve, and vh are the drift velocity of

electron and holes, respectively; and te, and th are the time of transit for

electrons and holes, respectively. We assumed that the excitation is made at a

location where te>th.

t

Ipho

te th Figure 2.5: Output current versus time for a constant illumination across the

depletion region.


For a constant illumination through the depletion region, the results

found in Eq. 2.7 must be integrated for all points inside the depletion region.

The result of this integration is shown in Figure 2.5. Here it is assumed that

ve>vh, which is the case for most of the semiconductors. The measured voltage

on the load resistance will depend on the capacitance of the photodetector and

the load resistance.

The junction resistance (Rj) is much higher than both the series (Rs) and

the load resistance (RL), so that it can be neglected in the calculations. The

junction capacitance and the series resistance form a low pass filter that limits

the bandwidth of the photodetector. The junction capacitance can be simply

found as Cj = εA/d, where ε is the electrical permittivity, A is the area of the

photodetector, and d is the width of the depletion region.

Ipho Rj

Rs

Cj

RL

Figure 2.6: Small signal equivalent circuit for photodetector.

The 3-dB roll-off frequency of the RC limited photodetector can be found as:

jsLRC CRR

f)(

121

+=

π

(2.8)

The other limit on the frequency response is the transit time of the carriers. The

3-dB roll-off frequency for transit limited case is:

dvf car

tr 45.0= (2.9)

where, vcar is the velocity of the carriers.


Another important characteristic of the photodetector is the quantum

efficiency or the responsivity. When a photon, with wavelength λ whose

energy is larger than the bandgap, is absorbed in the depletion region, an

electron hole pair is generated. These carriers are swept away by the electric

field. The number of electrons generated per incident photon is defined as the

quantum efficiency, which is expressed as [15]:

)/(/ν

ηhPqI

opt

p=

(2.10)

where, Ip is the photo-generated current, and Popt is the optical power at

frequency ν. The relation between the responsivity and the quantum efficiency

can be written as:

λη ℜ×=

24.1 (2.11)

where, λ is in microns, and responsivity (ℜ) is in amps per watt (A/W).

For classical single pass, vertical illuminated photodetectors, the

quantum efficiency is given by:

)1)(1( deR αη −−−= (2.12) where, R is the reflectivity of the front surface, α is the power absorption

coefficient, and d is the thickness of the active layer. So, to maximize the

quantum efficiency, the surface reflectivity must be minimized, and single pass

absorption must be maximized. Reflectivity can be minimized using anti-

reflection coatings, and single pass absorption can be maximized by increasing

layer thickness.

For the transit time limited photodetector, the thickness is small, and αd

is much smaller than one. In this case, the quantum efficiency can be

reformulated as:

dR αη )1( −= (2.13)


In this case, the bandwidth efficiency product can be obtained as:

cartr vRf αη )1(45.0 −=× (2.14) which is independent of the active layer thickness.

2.2 Resonant Cavity Enhancement For transit time limited photodetectors, the depletion region must be kept thin

enough to achieve high-speed operation. On the other hand, for high quantum

efficiency, the depletion layer must be sufficiently thick to absorb a high

fraction of incident light. To overcome this trade off between the response

speed and efficiency, a conventional photodetector with a thinner active layer

can be placed inside a Fabry-Perot micro-cavity. Thin active layer results in

lower transit time, but performance of the photodetector with the help of the

cycling of the optical power inside the cavity, is increased [10]. This kind of

enhancement of the quantum efficiency is called Resonant Cavity

Enhancement (RCE). The RCE effect was proposed in 1990 and was applied to

a broad range of detectors: Schottky [16, 17], p-i-n [18-20], avalanche [21, 22],

and MSM [50] photodiodes.

2.2.1 RCE Formulation and Optimization Figure 2.7 shows a generalized structure of an RCE photodetector. Although

metal coatings can be used as mirror, lossless distributed Bragg reflectors

(DBR) are used as mirrors of the micro-cavity, as the aim is to achieve

maximum efficiency. The active layer, where the absorption occurs, is placed

between these mirrors. L is the length of the cavity, and d is the thickness of

the active layer. The field reflection coefficients of the top and bottom

reflectors are and , where 11

ϕier − 22

ϕier −1ϕ and 2ϕ are phase shifts due to the

light penetration into the mirrors.


Substrate

d active layer

top mirror

bottom mirror

Ei

Ef

Eb

r1

r2

αext

αext

α

L1

L2

Figure 2.7: Generalized structure of a resonant cavity enhanced

photodetector.

Ei represents the electric field amplitude of the incident light, while Ef is

the forward traveling wave at z = 0, and Eb is the backward traveling wave at

z = L = L1+d+L2. In the cavity, Ef is composed of the transmitted wave from

the first mirrors and the reflected wave from the second mirror. Therefore, the

forward traveling wave, Ef, at z = 0 can be obtained in a self-consistent way:

fLiLLd

f EeerrEitE ext )2()(211

2121 ϕϕβαα ++−+−−+= (2.15)

where, β = 2πn/λ0 , α and αext are the absorption coefficients of the active and

cavity layers respectively. Solving for Ef gives us

Eieerr

tE LiLLdf ext )2()(21

121211 ϕϕβαα ++−+−−−

= (2.16)

and the backward traveling wave, Eb, at z = L can be expressed as:


fLi

LLd

b EeeerEext

)(2)(

22

2

21

ϕβαα

+−+−−

= (2.17)

The optical power inside the resonant cavity is proportional to the refractive

index of the medium and the square of the electric field amplitude.

nzEzP 2)()( ∝ (2.18) If we neglect the standing wave effect, the power absorbed in the active layer

in terms of the incident power is given by:

iLL

dLLL

a PerrLerr

eererPcc

cextext

αα

αααα

ϕϕβ 22212121

22

21

)()2cos(21)1)()(1( 21

−−

−−−−

+++−−+−

= (2.19)

where αc = (αext(L1+L2)+αd)/L. Under the assumption that all the

photogenerated carriers contribute to the current, η is the ratio of the absorbed

power to the incident optical power, i.e., η = Pa/Pi. Hence:

)1)(1()2cos(21)(

1212121

221

dLL

LLL

eReRRLeRR

eRecc

cextextα

αα

ααα

ϕϕβη −

−−

−−−

−−⎥⎥⎦

⎤

⎢⎢⎣

⎡

+++−+

=

(2.20)While designing the detector, the cavity layers are chosen such that all the light

is absorbed in the active layer αext ` α. The expression in the square braces is

called the enhancement, as it is the multiplier to the quantum efficiency of a

conventional photodiode. Enhancement can be rewritten as:

dd

d

eRRLeRReRtenhancemen

αα

α

ϕϕβ −−

−

+++−+

=212121

2

)2cos(21)1(

(2.21)

From this expression, it is seen that η is enhanced periodically at the resonant

wavelengths of the cavity, 2βL + 1ϕ + 2ϕ = 2πm (m = 1,2,3...). This term

introduces the wavelength selectivity of the RCE effect.


From the expression for the quantum efficiency (η), it is seen that three

parameters R1, R2, and αd effect η. R2, the reflectivity of the bottom mirror,

should be designed as high as possible. Otherwise, due to the transmission

from the bottom mirror, the performance of the detector decreases dramatically.

The other parameter is the reflectivity of the top mirror. When the quantum

efficiency is maximized with respect to R1, the following condition is obtained: deRR α2

21−= (2.22)

If top mirror reflectivity is low, the light is lost from the cavity in the form of

reflection. On the other hand, if it is high, incoming light cannot be coupled

into the cavity.

2.2.2 Standing Wave Effect While deriving the formula of the quantum efficiency, the spatial distribution

of the optical field inside the cavity was neglected. This spatial distribution

arises from the standing wave formed by the two counter propagating waves.

This is referred as standing wave effect (SWE). The SWE is conveniently

included in the formalism of η as an effective absorption constant, i.e., αeff =

SWE × α. The effective absorption constant αeff is the normalized integral of α

and the field intensity across the absorption region.

∫

∫= 2/

0

2

0

2

),(2

),()(1

λ

λλ

λαα

dzzE

dzzEzd

d

eff (2.23)

When detectors with thick active layers, which span several periods of the

standing wave, are considered, SWE can be neglected. For very thin active

layers, which are necessary for strained layer absorbers, SWE must be

considered.


2.3 Optical Design Our detector designs, like other RCE photodetectors, vertical cavity surface

emitting lasers (VCSEL), and distributed feedback lasers (DFB) are quite

complicated. The devices contain micro cavities, which are formed with low

loss mirrors. The reflection amplitude and phase from these mirrors are not

trivial. Also the optical properties of the materials used in the cavities are

wavelength dependent and this makes rather difficult to predict the optical field

in such multilayer devices, therefore a good simulation method is needed for

the analysis before the growth. Transfer matrix method (TMM) is commonly

used, which provides a simple and accurate technique to calculate electric and

magnetic field distributions inside the cavity.

2.3.1 Transfer Matrix Method

When a layer is modeled from an optical point of view, it is seen that it consists

of two elements; an interface where an abrupt refractive index difference

occurs, and a slab where refractive index is constant and extends for a certain

width. Refractive index is defined as the square root of the dielectric constant

of the medium; ε=n . Usually the dielectric constant is complex and the

imaginary part is due to the absorption in the medium, and refractive index can

be simplified as imagreal innn −= , where both and are real and

positive numbers.

realn imagn

The electric field at any point can be modeled as superposition of

forward and backward going waves as shown in Figure 2.8. When the Maxwell

equations are solved, the continuity of the electric and magnetic field at the

interface is obtained.


r12

t12

1 2

E1f

E1b

E2f

E2b

⎥⎦

⎤⎢⎣

⎡1

1112

12

12 rr

t

21

112

2nn

nt+

=

21

2112 nn

nnr+−

=

Figure 2.8: Electric field amplitudes at the interface.

The relation between the forward and backward field amplitudes before

and after the interface has the relation:

⎥⎦

⎤⎢⎣

⎡⋅⎥

⎦

⎤⎢⎣

⎡=⎥

⎦

⎤⎢⎣

⎡

b

f

b

f

EE

rr

tEE

2

2

12

12

121

1

111 (2.24)

where, t12=(2*n1)/(n1+n2), r12=(n1-n2)/(n1+n2). n1 and n2 are the refractive

indexes of the consecutive layers.

The electric field inside the layer can be found by propagation of plane

wave:

⎥⎦

⎤⎢⎣

⎡⋅⎥

⎦

⎤⎢⎣

⎡=⎥

⎦

⎤⎢⎣

⎡ −

b

fxik

xik

b

f

EE

ee

xExE

2

2

2

2

2

2

00

)()(

(2.25)

where, k=(2πn)/λ is the wave vector in the medium. When x is chosen as the

width of the layer (as shown in Figure 2.9), the electric field just at the left of

the next interface can be found.


x

2 2 2E2f

E2b

E2f(x)

E2b(x)

222 nkλπ

=

⎥⎦

⎤⎢⎣

⎡ −

xik

xik

ee

2

2

00

Figure 2.9: The propagation of wave inside the same medium.

If the two matrices given by eq. 2.24 and eq. 2.25 are combined, a transfer

matrix of the layer for the field amplitudes is given as:

⎥⎦

⎤⎢⎣

⎡=

−− mm

mm

iim

im

i

mm eer

eret

T δδ

δδ1 (2.26)

where, tm=(2 nm)/(nm+ nm+1), rm=(nm-nm+1)/(nm+ nm+1), and δm=kmdm. Cascading

these matrices for N layers, total matrix of the multiplayer structure can be

constructed as:

NNtotal TTTTTT 1210 −= (2.27) The relation between the electric field at electric field at the left and at the right

sides of our system as depicted in Figure 2.10 is given by:

⎥⎦

⎤⎢⎣

⎡⋅=⎥

⎦

⎤⎢⎣

⎡

fb

fftotal

ib

if

EE

TEE

(2.28)

For a detector structure shown in Figure 2.10, Efb is zero; hence Eff and

Eib can be solved in terms of Eif easily.


Eif

Eib

Eff

Efb

Figure 2.10: Electric field values before and after the structure.

When a measurement like reflectivity is made, the reflected power is

measured not the electric field. The same is valid for the transmission and

absorption. The power associated with a plane wave can be found by:

BES ×=µ1 (2.29)

and using the relation between the electric and magnetic field amplitudes for a

plane wave:

EkB ×=ω1 (2.30)

The power is proportional to the square of the electric field amplitude and the

refractive index of the medium. Then the reflectivity and transmittivity of the

structure are given by:

2

2

if

ib

E

ER = (2.31)

initif

finalff

nE

nET 2

2

= (2.32)


Calculating the total power entering and leaving that layer, we can find

absorption of this layer [23-25]. Also the same calculations apply for the

oblique incidence with small modifications. Sample TMM calculations are

shown at the appendix.

2.3.2 Distributed Bragg Reflectors As shown in the previous sections, the aim is to place the active layer inside a

cavity formed by two mirrors. Good designs require bottom mirrors with high

reflectivities. Although metals are good reflectors, their reflectivities (R≈95

can be achieved) are wavelength dependant, and semiconductor layers cannot

be grown on them.

Distributed Bragg reflectors (DBR's) are widely used in optoelectronic

applications such as detectors and semiconductor lasers. A DBR is a periodic

stack with two alternating quarter-wave thick materials with different optical

properties. Each pair consists of two layers with refractive indices n1 and n2,

and layer thickness of λc/4n1 and λc/4n2 respectively. λc is the central

wavelength of the mirror where reflectivity is maximum.

When the light is reflected from different interfaces of one pair, it is

observed that all the light is at the same phase resulting in a constructive

interference hence increasing the reflectivity. Applying TMM for a DBR at

center wavelength λc, results in a simple expression for the reflectivity as: 2

2

1

2

2

1

2

max

1

1

⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜

⎝

⎛

⎟⎟⎠

⎞⎜⎜⎝

⎛+

⎟⎟⎠

⎞⎜⎜⎝

⎛−

= N

N

nn

nn

R (2.33)

where N is the number of mirror pairs in the stack. For large values of N

reflectance approaches unity, and for a fixed N reflectivity increases as n2/n1


increases. Also width of the highly reflective wavelength span increases as the

index difference between the layers increases.

1000 1200 1400 1600 1800 20000

20

40

60

80

100

Ref

lect

ivity

(%)

Wavelength (nm)1000 1200 1400 1600 1800 2000

-4

-3

-2

-1

0

1

2

3

4

Pha

se D

iffer

ence

Wavelength (nm)

Figure 2.11: Reflection spectrum and phase difference of InP/Air DBR with 3 pairs.

The phase difference between the incidence and reflected wave is a

function of the wavelength. The reflectivity and the phase difference between

the incidence and reflected waves for InP/Air and InP/InAlGaAs DBRs are

shown in Figure 2.11 and Figure 2.12.

1000 1200 1400 1600 1800 20000

20

40

60

80

100

Ref

lect

ivity

(%)

Wavelength (nm)1000 1200 1400 1600 1800 2000

-4

-3

-2

-1

0

1

2

3

4

Pha

se D

iffer

ence

Wavelength (nm)

Figure 2.12: Reflection spectrum and phase difference of InP/InAlGaAs DBR with 40 pairs.


In the InP/Air DBR, the index difference is 2.167. Three pairs are

enough to get 99.6% reflectivity with this structure. If the high reflectivity band

of this structure is observed, it is seen that its FWHM is higher than 1000 nm.

While InP/InAlGaAs DBR has index difference of only 0.293. This DBR needs

40 pairs to achieve 99.5% reflectivity. Also due to lower index contrast,

FWHM is less than 100 nm.

With a small change in the design of the DBRs, the nature of these

mirrors can be changed. If the width of one of the layers (with a refractive

index of n1) increased from λc/4n1 to λc/2n1, the structure becomes transparent

around λc, while still reflecting other wavelengths. In this case, the multiplayer

stack functions as a notch filter. Such filters have been demonstrated at various

wavelengths on both silicon and GaAs based devices [26-28]. Figure 2.13

shows the transmission spectra of notch filter, which consists of Si and SiO2

layers operating around 1550 nm.

1400 1450 1500 1550 1600 1650 17000

20

40

60

80

100

Tran

smis

sion

(%)

Wavelength (nm)

Figure 2.13: Transmission spectrum of Si/SiO2 notch filter centered to 1550 nm.

Chapter 3

Fabrication and Characterization

This chapter presents the fabrication and characterization process that was

completed in Advanced Research Laboratories at Bilkent University. Section

3.1 explains the basic fabrication steps, while section 3.2 presents the

measurement setups.

3.1 Basic Fabrication Steps This section explains the details of fabrication steps completed in a Class-100

clean room environment. With the application of three or four of them, usually

one fabrication step of the photodetectors are completed.

3.1.1 Sample Cleaving The wafers that are used for the fabrication of photodetectors in this thesis are

usually quite expensive and hard to grow. That’s why we prefer working with

small samples cleaved from the wafer. Sample sizes are around 7×7 mm2, and

the mask is 6×6 mm2. We use a diamond tipped scriber-pen to define a line at

the back of the substrate. After defining the shape of the sample, the wafer can

be easily cleaved. The thicknesses of the wafer are usually 350 µm for 3-inch

wafers, and 600 µm for 4-inch wafers.

26

CHAPTER 3. FABRICATION AND CHARACTERIZATION 27

Figure 3.1: Photographs of some samples with fabricated photodetectors on them.

3.1.2 Sample Cleaning The samples are cleaned at every step. The residues from the previous step

must be cleaned. Three different solvents are used for the cleaning, namely

trichloroethane, aceton, and isopropanol alcohol. Samples are dipped into them

in order. The details of the cleaning are as follows:

• Samples are immersed into the boiling trichloroethane for 2 minutes.

• Then the samples are kept in acetone at room temperature for 5 minutes.

Acetone dissolves organic molecules, and the photoresist.

• Next samples are kept in boiling isopropanol alcohol for 2 minutes.

This alcohol cleans the residues from the previous step.

• Samples are rinsed in the de-ionized (DI) water flow and dried with

nitrogen gun.


• Although the sample is dried with nitrogen flow, there remains a very

thin layer of water at the surface. To evaporate this water, samples are

baked at 120 oC on the hat plate for 2 minutes. This bake is called

dehydration bake.

3.1.3 Photolithography Before the fabrication, the mask is prepared which contains the features to be

transformed onto the samples. The CAD program named Wavemaker is used

for the preparation of the digital file. The features for each step are placed on

6×6 mm2 area. These features are the photodetectors, test diodes that does not

contain interconnect metallization, ohmic test patterns, development marks that

are used to check the quality of the photolithography and development, and the

alignment marks that are used to align the mask with respect to the sample at

each step. One mask usually contains 25 dies, where each die has the patterns

for one lithography step. The size of the masks is 4×4 inch2 so that all the dies

can be used with the mask aligner. Figure 3.2 shows photographs of our masks.

Figure 3.3 has some plots from the digital mask file.

Figure 3.2: Photographs of the masks.


(a)

(b)

(c) (d)

Figure 3.3: Plots from the mask file (a) a photodiode with 100 µm diameter active area, (b) development marks used in the p+ ohmic step, (c) p+ ohmic

transmission lines and the development marks, and (d) alignment marks.

Photolithography starts with covering the surface of the sample with a

chemical, which contains polymer that is sensitive to ultraviolet radiation. This

chemical is called photoresist. The AZ-5214E type photoresist produced by

Clarient is used. To achieve a uniform thickness, the sample is spinned at 5000

rpm for 40 seconds. The thickness of the photoresist obtained after this spin is

around 1.35 µm. The thickness can be changed by changing the spin rate [29].

This photoresist is sensitive to UV radiation in the 310 – 420 nm spectral range.

The Karl-Suss MJB3 mask aligner is used for the alignment of the samples and

exposure. The aligner has a mercury lamp that has a high power emission

around 365 nm. Before the exposure, the sample is baked on a hot plate at


110oC for 55 seconds. This bake is called the soft-bake. At this point, the

process can be varied to get both positive and negative resist.

Figure 3.4: Plot of the mask containing large area photodetectors. All the fabrication steps are shown aligned to each other. The mask contains around

115 photodetectors.

• Normal Photolithography: After the soft-bake, the sample is

exposed with the mask with a total energy of 150 mJ/cm2. The

photoresist at the exposed area becomes soluble in the developer.

• Image-reversal Photolithography: After the pre-bake, the sample

is exposed with the mask with a total energy of 50 mJ/cm2. Then the

sample is baked at 110 oC for 2 minutes. The exposed photoresist

becomes inert after this step. Then the whole sample area is exposed

without a mask with a total energy of 150 mJ/cm2. Then the

unexposed area in the first exposure becomes soluble in the

developer. The explanation of this process is also shown in Figure

3.5.


substratephotoresist

soluble inert

soluble

(1) Exposure using an inverted mask (the exposed

areas finally remain

(2) The resist now would behave like an exposed

positive resist

(3) The reversal bake cross-links the exposed area, while the unexposed area remain

photo-active

(4) The Flood exposure(without a mask) …

… (5) makes the resist, which was not exposed in

the first step, soluble in developer

(6) After development, the areas exposed in the first

step now remain

Figure 3.5: Image-reversal photolithography (after reference [30]).

3.1.4 Development For the development of the exposed samples, AZ400K developer is used with a

1:4 (Developer:H2O) volume ratio. As the soluble regions of the resist are

etched by the developer, change in the color can be observed with naked eye.

When this color change stops, the sample is rinsed under DI water. After

drying the sample, the alignment and the development is checked. The

resolution patterns must be sharp, and for a good photolithography about 1 µm

resolution should be observable. The resolution of the image-reversal

photolithography is less than the normal photolithography. The AZ400K

developer contains metal ions and is based on potassium hydroxide. This

developer affects the semiconductor layers containing Aluminum.


3.1.5 Etch The etch process is used for transforming the defined patterns onto the

underlying metal, semiconductor or dielectric layers, so that the desired layer is

reached for the subsequent process. Etching is also used for cleaning and

thinning photoresist, removing damaged material, polishing, and removing

surface oxides. Etching is done in two different methods, either by using

chemical reactants present in aqueous solution (wet etch) or with the help of

ions and molecules present in a plasma (dry etch).

• Wet Etch: Chemical reactions that occur at the surface of the material

with the reactants present in the solution. For the semiconductors, the

etching is basically done by oxidation or reduction of the surface and

then removal of the soluble reaction product. The etch rate may be

limited by the transport of the reactants and the products. This type of

etches are diffusion-controlled etches. This type of etch is usually

anisotropic and agitation of the solution changes the etch rate and

profile. The etch is called reaction-limited etch, if the reaction has a

rate-limiting step. These type of etches can be isotropic, and the rate is

insensitive to agitation [31]. We bake the resist at 120 oC for 1 min, to

refrain the photoresist from the tension that occurs during the

development. This bake enhances the adhesion to the substrate at the

contours of the photoresist, which decreases the under-etch for deep

etches. Figure 3.6 (a) shows a wet etch profile of GaAs where the

under-etch is more than expected. Some of the etch recipes that are

used in our lab are listed in Table 3.1.


Figure 3.6

Table 3.1: Etch rat

Etchant

H3PO4 : H2O2 : H2O (1:3:HCl : H3PO4 (1:3)H3PO4 : H2O2 : H2O (1:1:Citric acid : H2O2 (1:1)HF : H2O2 : H2O (2:1:NH3 : H2O2 : H2O (2:1:

• Dry Etch: Dry e

reactions or energeti

plasma etching, reac

(RIBE), sputter etch

the bombardment of

etching is made mec

(a)

: SEM picture wet etch profiles.

es of some recipes used in our processes.

Etched Layer Etch R(nm/s

40) InGaAs, InAlAs 4.5, 5 InP 12.5) InAs 12. InSb ~ 0.4 -100) GaSb 30.75) GaAs, Al0.2Ga0.8As 5.0, 8

tching techniques use plasma-driven

c ion beams. Some of dry etching techn

tive ion etching (RIE), reactive ion-beam

ing, and ion milling. Plasma etch occurs

the surface with highly energetic ions w

hanically, or by the chemical reactions bet

(b)

ate ec) .0

5 5 0.5 0 .1

chemical

iques are

etching

either by

here the

ween the


gases and the surface. In the later, volatile products are generated which

are pumped out of the reaction chamber. Changing the gas pressure and

ion energy can control degree of isotropy. Usually higher gas pressure

results isotropic etch. Ultra high vacuum (UHV) RIE machine is used

for dry etching. With appropriate gases, dielectrics, semiconductors,

resist, even metals can be etched with this system [32-34]. Currently

CCl2F2, SF6, CHF3, O2, CF4, and H2 gases are available in our system.

Some of the recipes are listed in Table 3.2.

Table 3.2: Etch rates of some recipes used in RIE etch.

Plasma Condition Etch Results

RF Power : 51 W Pressure : 7 µBar O2 : 20 sccm Self Bias : 350 V

Hardened Photoresist : 70 nm/min

RF Power : 51 W Pressure : 4 µBar CCl2F2 : 20 sccm Self Bias : 360 V

GaAs : 100 nm/min Al0.3Ga0.7As : 30 nm/min Photoresist : 10 nm/min


InP : 50 nm/min InGaAs : 55 nm/min

Photoresist : 12 nm/min

RF Power : 200 W Pressure : 15 µBar CH4 : 10 sccm H2 : 20 sccm O2 : 0 sccm Self Bias : 570 V

InP : 36 nm/min (CH4 currently not available)

(Damages the photoresist) (Polymer deposition on the mask)


RF Power : 69 W Pressure : 20 µBar CHF3 : 60 sccm O2 : 5 sccm Self Bias : 350 V

Si : 12.5 nm/min Si3N4 : 70 nm/min SiO2 : 1.1 nm/min

Photoresist : 32 nm/min


GaAs : 300 nm/min InAs : 60 nm/min Si3N4 : 40 nm/min SiO2 : 25 nm/min ITO : 20 nm/min

Photoresist : 45 nm/min GaN : 31 nm/min

Al0.38Ga0.62N : 13 nm/min Hardened Photoresist : 22 nm/min

RF Power : 150 W Pressure : 10 µBar CHF3 : 60 sccm O2 : 5 sccm Self Bias : 550 V

SiO2 : 17 nm/min Photoresist : 42 nm/min

While most of the III-V semiconductors materials can be etched using a

wet etch techniques, GaN and related materials cannot be etched easily in this

way. We use only RIE with CCl2F2 gas to etch GaN and other wide band-gap

materials [35].

3.1.6 Metallization Metals are deposited onto the sample for ohmic, Schottky, or interconnect

metallization. Metals and dielectric coatings are deposited inside ultra high

vacuum LE590 box coater, using thermal evaporation or radio frequency (RF)

sputtering technique. In both methods, the pressure of the process chamber is

lowered to 5×10-6 mBar. The deposition rate and the total thickness of the film

are monitored during the process.


We usually pattern the metal layer using a technique called lift-off.

Metal is deposited on patterned photoresist. Deposited metals stick to the

surface of the sample at the openings. The metals on the photoresist are lifted-

off later. For the thick metallization, image-reversal photolithography is

preferred. The negative slope of the photoresist enables a discontinuity

between the metal at the openings and the photoresist.

• Thermal Evaporation: Pallets or powder of metals are placed inside

tungsten boats. When we pass current though the boat, metals start

evaporating. By changing the current, we can control the evaporation

rate of the metals. Hot metal atoms from the boats spread inside the

vacuum and stick to the sample and cold surfaces. Gold (Au),

Germanium (Ge), Nickel (Ni), Titanium (Ti), and Aluminum (Al) are

deposited with this method in our lab.

• RF sputtering: Samples are placed on top of the target. The target

contains the material to be deposited on the sample. With the help of

argon and the RF power, argon atoms break bonds of the target. The

ions from the target move up to the sample. If enough ions accumulate

at the surface, they join and nucleate. After forming islands on the

surface, they combine and for a layer of film. Indium-tin-oxide (ITO) is

deposited with this method.

3.1.7 Lift-off After the metallization, the metals deposited on top of the photoresist must be

removed from the sample. Samples are soaked into the acetone. As acetone

dissolves the photoresist, the metals on them also are lifted-off from the sample.

This process is quite easy if these metals have no connection with the

ones at the openings. Otherwise ultrasound must be used to agitate the sample


and break the contact between these two metal regions. It is easier to use the

lift-off technique if thermal evaporation is used compared to the sputter

technique, as thermal evaporation does not produce a good step coverage on

the sidewalls of the photoresist.

3.1.8 Thermal Annealing Rapid thermal annealing (RTA) is used to heat the samples very rapidly to very

high temperatures (as high as 1400 oC) to form ohmic contacts. Flash lamps are

embedded into the RTA device as the heat source. Samples are heated under

the radiation of these flash lamps for a certain time. The environment contains

forming gas, which is a mixture of hydrogen (5%) and nitrogen (95%).

The contact between the metal and the semiconductor surface has a

barrier called Schottky barrier, hence this is a rectifying contact. The electrons

can flow from one region to the other either by thermionic emission over the

barrier or by tunneling though it. Most ohmic contacts are achieved by

decreasing the width of the potential barrier.

As the doping of the semiconductor increases, the width of the barrier

decreases, and the tunneling can begin without a resistance, i.e., resulting in a

good ohmic contact. The high doping can be achieved by alloying the metal

with the semiconductor, which is possible with the thermal treatment. The

alloy temperature and time has an important effect on the resistance of the

contact. Increasing the temperature to the optimum value decreases the contact

resistance down to the minimum achievable value. Further increase of the

temperature usually increases both the resistance and the morphology of the

contact [37]. The optimum annealing conditions of each ohmic contact must be

determined for every different wafer.


The resistance of the contact is determined using a series of ohmic

contacts with varying distance between them. The crossection of such contacts

are shown in Figure 3.7 (a). The total resistance can be modeled as two contact

resistances (Rc) and the sheet resistance (Rs) of the semiconductor as shown on

Figure 3.7 (b). By plotting the resistance versus distance between contacts (L)

graph, we determine the contact resistance (half of the y-intercept) and the

sheet resistance of the semiconductor layer (slope of the line times the width of

the ohmic contact). Figure 3.7 (c) shows the series of ohmic contacts used for

this measurement.

Figure 3.7: (a) The current flow

tran

.1.9 Dielectric Deposition

model used for the ohmicsmission line (center) is u

alignment marks (up) and t

)

L

Semi-Insulating Layer

L

Semi-Insulating Layer

RC RC

RS

3Dielectric layers are used for m

and photodetectors. The dielec

(a

)

)
(b between the contacts. (b) The schem
-insulator-metal capacitors, CMO

contact characterization. (c) The p+ osed for ohmic quality measurements.

he development marks (bottom) can b

etal

tric coatings in our process are used f

(c

atics of the

e

S devices,

hmic Also the seen.

or surface


passivation, anti-reflection coating, and top DBRs. These dielectric layers can

be deposited by RF sputtering or plasma enhanced chemical vapor deposition

(PECVD) technique in our lab. The repeatability of the first method isn’t very

good. The optical and mechanical properties of the films can fluctuate very

much through the sample. Instead, PECVD method is preferred, which has

high uniformity and repeatability. Also the optical and mechanical properties

of the films can be controlled with the flow rates of the gases and process

temperature.

Our device is a computer controlled Plasmalab 8510C reactor. It has a

chamber that contains a parallel plate for the application of RF (at 13.56 MHz),

heater, temperature controller, gas flow controllers, and pressure controller.

With the application of RF signal, we produce chemically active species inside

the plasma and reaction takes place at the hot surface of the sample. The gases

used in our lab are silane (2% SiH4:98% N2), ammonia (NH3), and nitrious

oxide (N2O). The deposited films are amorphous, and byproducts are pumped

out of the process chamber. The overall reactions for the deposition of SiO2

and Si3N4 films are as follows:

24334 1243 HNSiNHSiH +→+

and 22224 222 NHSiOONSiH ++→+

It is better to use lower RF powers (10–50W) and moderate

temperatures (200–350 oC) to get a good quality films. The refractive index of

the films can be controlled by changing the flow rates of the gases [36]. In our

process, we try to get high index nitride layers for the antireflection coatings.

Increasing the silane flow rate can increase the refractive index. But increasing

it to high values results in films that can’t be etched easily with diluted HF. We

usually prefer 20 W RF power during the growth. Our process conditions and


the properties of the resulting films are given in Table 3.3. Figure 3.8 contains

pictures of the devices used during these process steps.

(a)

(b)

(c)

(d)

(e)

(f)

Figure 3.8: Pictures of equipments: (a) mask aligner, (b) surface profilometer,

(c) thermal evaporator and RF sputter, (d) PECVD, (e) RIE, (f) RTA


Table 3.3: Growth conditions and properties of dielectric films.

Plasma Condition Film Properties RF Power =10 W RF Power =20 W RF Power =50 W

Silane : 180 sccm NH3 : 45 sccm Pressure: 1000 mTorr Temp : 250 oC

Rate: 10.7nm/minIndex: 1.72

Rate: 12.8nm/min Index: 1.76


Silane : 180 sccm N2O : 710 sccm Pressure: 1000 mTorr Temp : 250 oC


Rate: 40.6nm/min Index: 1.46


3.2 Measurement Setups This section briefly explains the measurement setups and techniques in our

measurement lab in Class-10000 environment.

3.2.1 Reflection and Transmission Measurements The reflectivity of the grown wafer is the first data that can be used to

understand the layer properties of our detectors. Also the reflection and

transmission results of dielectric films give us the optical properties of these

films. A fiber optic based setup is used for these measurements. The light

source and the detector change for different wavelength range.

For the measurements at the ultraviolet (UV) and visible range,

deuterium-tungsten light source is used, and the light is coupled to reflection

probe. The reflected or the transmitted light is again coupled to another fiber,

which delivers the light to the spectrometer located in one of the PCI bus of the

computer.


For the infrared region we use two different setups. First consists of a

halogen lamp whose light passes through a monochromator and coupled to the

fiber arrangement explained before. Then the optical power is measured with

optical powermeter. The other setup consists of the halogen light source and

the fiber probes. But the reflected of transmitted light is measure with optical

spectrum analyzer. The setup used for the infrared measurement is shown in

Figure 3.9.

(a) (b) Figure 3.9: (a) The reflectivity measurement setup. The light source (blue) the

fiber probe and the optical spectrum analyzer can be seen. (b) The reflectivity of the InGaAs based photodetector wafer.

3.2.2 Current Voltage Measurements We make the on-wafer measurements with the help of probe station (shown in

Figure 3.10(a)), where various probes including DC contact probes, fiber

holder, and microwave connection probes can be placed. These probes are used

for the electrical and optical characterization.

For the current-voltage characterization we use computer controlled

HP4142B DC Source monitor unit. Using the computer, either current or

voltage inputs are provided, and the corresponding characteristics are measured

with the same module. This setup is also used for the characterization of the


ohmic contacts with the help of accurate voltage measurement ports. Figure

3.11 shows the unit and the computer control.

(a) (b) Figure 3.10: (a) The probe station. With the help of objectives, we can make

contact to the pads of the photodetectors. (b) The sample on the probe station, where two DC probes touch the pads, while the active are is illuminated with

the fiber.

) (a) (b)

Figure 3.11: HP4142B and the computer control.

3.2.3 Quantum Efficiency Measurements The responsivity characterization can be made with two measurement setups.

First setup consists of a wide spectrum light source and a monochromator for

wavelength selection. The light is chopped and coupled to a multimode fiber.

)
(c
(c


Using the probe station and the fiber probe, the active area of the detector is

illuminated. The detectors are biased with a DC power supply, and the

photocurrent is measured with a lock-in amplifier. The data is collected using a

computer equipped with a GPIB control. This setup has a wide spectrum range,

starting from UV to mid mid-IR wavelengths. The schematic diagram and the

pictures of this are shown in Figure 3.12.

Probe Station

Lock-in Amplifier(SRS830 Lock-In Amplifier)

Monochromator(CVI 1/4m Digikröm)

Chopper(NewFocus 3501 Optical Chopper)

fiber

Power Supply(HP E3631A Power

Supply)

Optical Powermeter(Newport 1830-C Optical Power

Meter )

GPIB Controller

(a)

(b)

(c)

(d) Figure 3.12: Quantum efficiency measurement setup with the monochromator.

The second setup consists of a tunable laser as the light source. It

covers the spectrum from 1500 to 1620 nm. The maximum optical power

output is a function of wavelength. The output wavelength and power is

controlled with a computer and the laser is coupled to a single mode fiber. The

light is delivered to the active area of detectors as described before. The


detectors are biased with DC power, supply and the photocurrent is measured

directly with multimeter. The data is then collected by a computer. The

schematic diagram and the picture of the setup are shown in Figure 3.13.

Probe Station

Tunable Laser(Photonetics OSICS-ECL)

DC power supply(HP E3631A)

Optical powermeter(Newport 1830-C)

Multimeter(HP 34401A)

3dB coupler

GPIB Controller

(a)

(b) Figure 3.13: Quantum efficiency measurement setup with the tunable laser.

3.2.4 High Speed Measurements High speed measurements are done with a laser that can generate optical pulses

with femtosecond widths. The pulses are generated using erbium doped fiber

laser, and coupled to single mode fiber. The detectors are biased with DC

power supply using bias-tee, and the temporal response of the detectors is

measured with a 50 GHz sampling scope. Although the scope is 50 GHz, the

bias-tee can operate up to 40 GHz, and this setup is limited with this frequency.

The response of the detector is stored as a function of bias voltage and the light

intensity. The schematics of the setup are shown in Figure 3.12 along with its

pictures.


Probe Station Sampling Scope(HP 54750A Digital Scope)

FemtosecondPulsed Laser

(Calmar Erbium Doped

Fiber Laser)

Amplifier

(Calmar EDFA-02)

Optical SpectrumAnalyzer

(HP 86140A)

fiber

Bias Tee

DC power supply(HP E3631A)

3dB coupler

Figure 3.14: High Speed measurement setup.

Chapter 4

InGaAs p-i-n Photodetector

This chapter presents our effort to fabricate high speed and high efficiency

RCE photodetector operating around 1550 nm wavelength. Photodetectors

based on InP lattice matched materials are most commonly used in the optical

fiber communication systems, eye-safe rangefinders, free space communication

systems, and laser pulse shaping. These applications require fast photodetectors

that can operate in tens of gigahertz range. Mixed materials lattice matched to

InP makes possible to make heterojunctions where this is impossible for

Silicon and Germanium. These materials are formed by mixing Group-III and

Group-V elements to form ternary and quaternary compounds. Figure 4.1

shows the lattice parameters and energy gap values of common III-V

compound semiconductors.

The photodetector performance is measured by the bandwidth-

efficiency product (BWE) and is limited for conventional vertically illuminated

photodetectors (VPDs) due to the bandwidth-efficiency trade-off [10]. This

tradeoff is due to the fact that the quantum efficiency and bandwidth of a

conventional VPD have inverse dependencies on the photoabsorption layer

thickness.

47

CHAPTER 4. INGAAS P-I-N PHOTODETECTOR 48

Figure 4.1: Lattice parameter vs. energy gap of III-V compounds used with

InP and GaAs based devices (after reference [61]).

One detection scheme to overcome this limitation is edge-coupled

photodiodes. This scheme has been used to achieve very high speed metal-

semiconductor-metal (MSM) [38] or p-i-n waveguide photodiodes with

bandwidths above 500 GHz [39], distributed MSM photodetectors with 78

GHz bandwidth [40], avalanche photodiodes with 120 GHz gain-bandwidth

product [41], traveling-wave photodetectors with high output current [42] or

115 GHz bandwidth [43], and 110 GHz 50%-efficiency mushroom-mesa


waveguide photodetectors [44]. The disadvantages of edge-illuminated

detectors are complex fabrication and integration along with difficult light

coupling.

The ease of fabrication, integration, and optical coupling makes the

resonant cavity enhanced (RCE) PDs attractive for high-performance

photodetection [10, 45-27, 20]. The incident photons, which are at the

resonance wavelength of the detector cavity, are recycled, so that the quantum

efficiency (QE) is enhanced at this wavelength. Therefore, by using RCE-PD

with a thin active layer, high efficiency values can be achieved without

lowering the detector bandwidth [48, 49]. Using the RCE structure, InGaAs

based MSM PDs with 77% QE and 10 GHz bandwidth at 1.3 µm wavelength

[50], Schotttky PDs with 55% QE and 22.5 GHz bandwidth [51], p-i-n [52],

and avalanche PDs with ~70% QE and 24 GHz at unity gain [54] have been

reported by other researchers. A comparison of this work and the other

research results are given in Table 4.1.

4.1 Design In0.53Ga0.47As was chosen to be the active layer of the detector. This ternary

compound is lattice matched to InP, and it can be easily grown by molecular

beam epitaxy (MBE) or metal-organic chemical vapor deposition (MOCVD)

techniques. In0.53Ga0.47As is direct band-gap material with the band-gap energy

of 0.75 eV, which corresponds to 1650 nm cut-off wavelength. InP based

photodetectors are superior to Ge photodetectors as, In0.53Ga0.47As is capable of

detecting optical signals transmitted at non-dispersion wavelength (1300 nm)

and minimum-loss wavelength (1550 nm) of the silica based fibers used in the


Tabl

e 4.

1: C

ompa

rison

of t

his w

ork

with

the

prev

ious

rese

arch

RC

E Ill

um.

Type

λ

(µm

) Q

.E.

Ban

dw.

(GH

z)

BEP

(GH

z)

I max

(mA

) C

omm

ent

Ref

.

√ ve

rt.

APD

1.

58

0.66

24

.0

16.0

[5

4]

√ ve

rt.

Pin

& S

ch 1

.57

0.20

Bot

tom

illu

min

ated

, met

al to

p m

irror

[5

2]

√ ve

rt.

Scho

ttky

1.55

0.

55

22.5

14

.6

IT

O S

chot

tky

with

subs

trate

rem

ovel

[5

1]

× ve

rt.

Scho

ttky

1.55

0.

46

25.0

11

.5

IT

O S

chot

tky

[53]

√ ve

rt.

pin

0.98

120

Ver

y sm

all a

rea,

pow

er c

oupl

ing

is a

pro

blem

[3

9]

√ ve

rt.

pin

1.47

0.

86

W

afer

fuse

d to

AlA

s/G

aAs m

irror

[5

5]

√ ve

rt.

MSM

1.

30

0.77

10

.0

7.7

[50]

× ve

rt.

MSM

1.

55

0.13

38

.0

4.9

In

tegr

ated

to re

ceiv

er w

ith o

ptim

ized

vav

egui

de [

56]

× ve

rt.

MSM

1.

30

0.25

16

.0

4.0

[57]

× ve

rt.

MSM

1.

30

0.92

4.

0 3.

7

Bot

tom

illu

min

ated

[5

8]

× W

G

APD

1.

55

0.23

27

.0

6.2

[41]

× W

G

pin

1.55

0.

50

110

55

W

aveg

uide

spec

ially

shap

ed fo

r hig

h re

spon

s. [4

4]

× W

G

MSM

1.

55

0.33

18

.0

6.0

12

Vel

ocity

mat

ched

des

ign

[59]

× W

G

pin

1.55

0.

38

35

13.1

45

.3

Vel

ocity

mat

ched

des

ign

[60]

√ ve

rt.

pin

1.57

0.

66

28

18.5

6.

2 Th

is w

ork


modern communication systems. In0.53Ga0.47As based detector structures have a

few disadvantages when compared to GaAs based structures:

1. In0.53Ga0.47As has an absorption coefficient 64% lower than GaAs as

shown on Table 4.2 [62]. Also the drift velocities of electron and holes

are around 80% lower than the carriers in GaAs as shown on Figure 4.2.

When both of these properties are combined, the bandwidth efficiency

product of In0.53Ga0.47As based photodetectors is 50% lower than GaAs

based detectors.

Table 4.2: Optical properties of various III-V semiconductor compounds.

Energy band-gap Eg (eV)

Cut-off wavelength

λg (nm)

Refractive Index

Absorption Coefficient

n (λ=1550nm) α (λ=1550nm) (cm-1)

InP 1.35 920 3.167 0 In0.52Al0.48As 1.45 855 3.200 0 In0.53Ga0.47As 0.75 1650 3.589 0.81×104

In0.53Al0.13Ga0.34As 0.91 1360 3.428 0 In0.66Ga0.34As0.73P0.27 0.86 1440 3.425 < 40

n (λ=820nm) α (λ=820nm) (cm-1)

GaAs 1.42 870 3.670 1.27×104

AlAs 2.18 570 2.990 0 Al0.20Ga0.80As 1.61 770 3.460 0

2. Band discontinuities are the results of heterojunctions. Although this is

a desired property for lasers, modulators, and quantum well devices

where the charges are confined inside the structure, we try to avoid this

in the high speed photodetectors. These discontinuities result in charge

trapping, and degrade the performance in terms of speed and


responsivity. The semiconductors used in the In0.53Ga0.47As based

photodetectors have larger energy band discontinuities compared to

GaAs based ones. Graded regions are usually formed to get a smoother

variation. Table 4.3 lists properties of a few junctions commonly used

in the InP and GaAs based photodetectors.

Table 4.3: Band discontinuities of some heterostructure systems.

Energy band-gap difference ∆Eg (eV)

Conduction band

discontinuity ∆Ec (eV)

Valance band

discontinuity ∆Ev (eV)

InP / In0.53Ga0.47As 0.60 0.25 0.35 In0.52Al0.48As / In0.53Ga0.47As 0.71 0.52 0.19 In0.52Al0.48As / InP 0.11 0.35 -0.24

Figure 4.2: Charge drift velocity as a function of electric filed strength for

various semiconductors (after reference [61]).


Alloy Matching Condition y

AlxGa1-xPySb1-y (0.227+0.040x)/(0.645+0.040x) AlxGa1-xAsySb1-y (0.227+0.040x)/(0.443+0.033x) AlxGayIn1-x-yAs (0.189-0.398x)/0.405 AlxIn1-xPyAs1-y (0.189-0.398x)/(0.189+0.020x) AlxIn1-xPySb1-y (0.610-0.343x)/(0.610+0.075x)

AlxIn1-xAsySb1-y (0.610-0.343x)/(0.421+0.055x) AlPxAsySb1-x-y (0.267-0.685x)/0.058 GaxIn1-xPyAs1-y (0.189-0.405x)/(0.189+0.013x) GaxIn1-xPySb1-y (0.610-0.383x)/(0.610+0.035x)

GaxIn1-xAsySb1-y (0.610-0.383x)/(0.421+0.022x) GaPxAsySb1-x-y (0.227-0.645x)/0.443

(a)

Alloy Energy gap (eV) Ternaries AlxIn1-xAs

Eg(Γ) = 0.37+1.91x+0.74x2

Eg(X) = 1.82+0.4x GaxIn1-xAs Eg(Γ) = 0.324+0.7x+0.4x2

GaAs1-xSbx Eg(Γ) = 1.43-1.9x+1.2x2

InAs1-xPx Eg(Γ) = 0.356+0.675x+0.32x2

Quaternaries GaxIn1-xAsyP1-y

Eg(Γ) = 1.35+0.668x-1.068y+0.758x2+0.078y2-0.069xy-0.322x2y+0.03xy2

Eg(Γ) = 1.35-0.775y+0.149y2

(x=0.47y : lattice matched to InP) AlxGayIn1-x-yAs Eg(Γ) = 0.36+2.093x+0.629y+0.577x2+0.436y2+

1.013xy-2.0xy(1-x-y) Eg(Γ) = 0.764+0.495z+0.203z2(0.98x+y=0.47, x=0.48z : lattice matched to InP)

(b)

Figure 4.3: (a) Lattice match conditions of quaternary compounds to InP (after reference [68]). (b) Energy gap of III-V compounds (after reference [69]).

3. Another big issue is the growth of highly reflective DBR for the RCE

photodetectors operating in the infrared region. The semiconductor


compounds lattice matched to InP have lower index contrast compared

to materials lattice matched to GaAs. AlAs / Al0.20Ga0.80As pair is a

perfect candidate for DBR centered around 800 nm. The index contrast

of ∆n = 0.47 makes high reflective mirrors with fewer pairs of

semiconductor layers. Unfortunately the index contrast available with

the InP matched systems is substantially smaller than between GaAs

and AlAs. The maximum available contrast ∆n = 0.40 is between InP

and In0.53Ga0.47As, and this DBR is not suitable for operation at 1550

nm due to absorption in the In0.53Ga0.47As layer. Thus some other

ternary or quaternary compounds with cut-off wavelength smaller than

1550 nm are used in high quality DBRs. These compounds are obtained

with adding Al or P to In0.53Ga0.47As. Figure 4.3(a) gives a summary of

quaternary compounds lattice matched to InP. The low index materials

are AlAs0.56Sb0.44, InP, or In0.52Al0.48As with refractive indexes 3.07,

3.17, and 3.20 respectively. While the high index materials can be

AlGaAsSb, AlGaInAs, or InGaAsP with the cut-off wavelength around

1450 nm where the refractive indexes of these materials hover around

3.45 [63-66]. The available index contrast ∆n is around 0.25, which is

half of the index contrast available in GaAs based materials. Due to this

lower contrast, the number of DBR pairs is around 36 to achieve

reflectivity higher than 99% [67].

The design started with the bottom DBR. Although higher index contrast can

be achieved with antimony (Sb) containing compounds, they are not easy to

grow. Other common alternatives are InAlAs/InAlGaAs and InP/InGaAsP

systems, which are commonly grown with MBE or MOCVD reactors. We


decided on the Al based DBR due to an MBE system available during the

design.

In0.53Al0.13Ga0.34As was chosen as the high reflective index material,

and it corresponds to χ = 0.27 for the definition given by ref [70]. The optical

properties of this semiconductor compound can be obtained by the equation

Eq.(4.1):

22

22)(

CBAn−

+=λ

λλ (4.1)

where A=9.255, B=1.599, and C=928.3nm. The refractive index of this

compound is 3.428 at 1550 nm while In0.52Al0.48As has refractive index of

3.200 as shown on Figure 4.4. The index contrast is 0.23 for this DBR pair.

Figure 4.4: Refractive index of (In0.52Al0.48As)χ(In0.53Ga0.47As)1-χ as a function of wavelength for different values of χ (after reference [70]).

The energy bandgap of In0.53Al0.13Ga0.34As was calculated according to the

formula given in Figure 4.3 (b). It was found that the energy bandgap was

0.912 eV, which corresponds to the cutoff wavelength of 1360 nm that is well

below 1550 nm. The TMM simulations show that the bottom DBR should have


a maximum reflectivity of nearly 90% at 1548 nm. Calculated spectral

reflectivity of the DBR is shown in Figure 4.5.

1400 1450 1500 1550 1600 1650 17000

20

40

60

80

100

FWHM of DBRaround 96 nm

Ref

lect

ivity

(%)

Wavelength (V)

max reflectivity of 89%at 1548nm wavelength

Figure 4.5: Spectral reflectivity of the bottom DBR calculated using TMM.

On top of the bottom DBR, the cavity started with a spacer

In0.52Al0.48As layer with the thickness chosen to fix the resonance around 1550

nm. This layer was also used as the insulating layer that the interconnect metals

are deposited. On top of this layer, a silicon doped In0.52Al0.48As layer was used

to make the n+ ohmic contacts. Before the active lightly n type doped

In0.53Ga0.47As layer a thin In0.52Al0.48As layer and a digital grading was used.

The digital grading was formed by varying the thickness of In0.52Al0.48As and

In0.53Ga0.47As layers, which effectively changes the energy bandgap of the

layers seen by the charges. On top of the active layer a similar grading was

placed for the holes. P+ ohmic contact layer was also transparent In0.52Al0.48As

layer doped with beryllium. The thickness of this layer was chosen to be thick


enough to tune one resonance of the cavity by post-process etching. A thin

highly doped In0.53Ga0.47As layer was used at the top of the structure, which

makes a lower resistive p-type ohmic contact possible. This layer also prevents

the oxidation of the wafer before the fabrication. The total thickness of the

photodetector exceeds 7.1 micron. The thickness and the doping information of

the layers are given in Table 4.4. A similar structure without the bottom DBR

was also designed to see the effect of the RCE effect directly by comparing the

quantum efficiency measurements.

Table 4.4: Epitaxial structure of the InGaAs based pin detector wafer.

Material Thickness (nm)

Doping Type

Dopant Concant. (cm-3)

In0.53Ga0.47As 30 p+ Be 1×1019

Digital Grading 30 p+ Be 1×1019

In0.52Al0.48As 210 p+ Be 1×1019

In0.52Al0.48As 50 n- Si 1×1016

Digital Grading 30 n- Si 1×1016

In0.53Ga0.47As 300 n- Si 1×1016

Digital Grading 30 n- Si 1×1016

In0.52Al0.48As 60 n- Si 1×1016

In0.52Al0.48As 300 n+ Si 3×1018

In0.52Al0.48As 240 S.I. none none In0.52Al0.48As 121 S.I. none none

25 pairs In0.53Al0.13Ga0.34As 112 S.I. none none 600 µm thick Semi-Insulating InP Substrate

The TMM based calculations showed that the quantum efficiency of the

photodetector was 74% while the reflectivity and transmittivity of the

photodetector were 7% and 14% respectively. Some of the power was also lost

in the bottom DBR. As a result of the low reflectivity of the bottom DBR, we

saw that the enhancement of quantum efficiency with the deposition of the top


DBR would be small. This is mainly due to more power loss from the bottom

of the detector. According to these results, we decided not to deposit silicon

nitride - silicon dioxide top mirror layer, instead use the reflectivity of the top

semiconductor layer. The calculations used during the design of the detector

are shown on appendix.

We also made calculations to predict the frequency response of the

photodetector. The high speed measurements were usually done under 7 to 9

volts of reverse bias. For our structure, this corresponded to electric field

strength of 20×103 V/cm. The drift velocities of the charged carriers were

around 8×106 cm/s and 5×106 cm/s for electrons and holes respectively.

Corresponding transit times were 7.6 ps and 4.9 ps. For a 14×14 µm2 area

photodetector, the calculated 3-dB bandwidth was 45 GHz for 50 Ω load

resistance [71]. As the sum of load and serial resistance increased, the 3-dB

bandwidth of the photodetector decreased. A sample calculation of the

frequency response of these photodetectors is given on appendix.

With the above electrical and optical properties, the calculated

bandwidth efficiency product was 33 GHz. The enhancement was more than 3

times compared to a similar single-pass structure. The calculations show that

the quantum efficiency of the single pass structure was 22% at 1550 nm.

4.2 Fabrication The structure was grown by MBE method by Quantum Epitaxial Designs, Inc.

in USA. Initial visual inspection showed that the wafer of the resonant

structure was not as shinny as the single-pass structure wafer. This was most

probably due to the growth of the quaternary compound used in the DBR. Our


measurements of the completed detectors fabricated from both wafers showed

no difference in terms of electrical characteristics.

Before the fabrication, we measured the spectral reflectivity of the

wafer which gives useful information about the thicknesses of the DBR and

cavity layers. Our reflectivity measurement and TMM calculations showed that

the bottom DBR had grown %3 to %4 thicker than our design. This shifted the

center wavelength of the DBR to around 1600 nm. The measurement and

calculation results for a region on the wafer where the shift was %4 are shown

on Figure 4.6. The lower and upper edge of highly reflective spectrum was

1550 nm and 1690 nm.

1450 1500 1550 1600 1650 1700 17500

20

40

60

80

upper edgeof the DBR

Measurement Fitted Simulation

Ref

lect

ivity

(%)

Wavelength (nm)

lower edgeof the DBR

Figure 4.6: The measurement (solid line) and calculation (dotted line) results

for a region on the wafer where the shift was %4.

The fabrication was made with two different masks. The first mask had

5 lithography steps having larger area photodiodes. These photodiodes were


used during quantum efficiency measurements, as it was easier to couple all the

light from cleaved single mode (~9 µm core diameter) and multi mode (~85

µm core diameter) fibers to the active area of the photodetectors. The area of

the photodetectors was ranging from 30 µm to 200 µm in diameter. The second

mask had 7 lithography steps having smaller area photodiodes. These

photodiodes were used during high speed measurements, not to limit the

responses of the photodetectors with the capacitance of the detector. Although

we used cleaved fiber during the high speed measurements, lensed fiber can be

used to couple the entire laser power into the photodetector. The area of the

photodetectors was ranging from 5×5 µm2 to 20×20 µm2, as well as a few

150×150 µm2. These photodetectors also had airbridge for lower parasitic

capacitance and integrated biasing capacitors for providing constant bias

during the excitation of the detector with high powered pulses.

4.2.1 Large Area Photodetector Fabrication 1. n+ Ohmic Contact Formation: After the normal photolithography, the

layers were etched down to the middle of the n+ doped InAlAs layer.

Wet (chemical) etch was used during this step. A phosphoric acid based

solution was used for etching both InGaAs and InAlAs layers. The

composition was chosen as H3PO4:H2O2:H2O (1:3:40) which had the

etch rate of 4.5 nm/sec for both of the materials. Increasing the ratio of

H2O2 to H3PO4, the etch rate of Al containing layers increased which

caused high under etch. A list of available etchants used with the InP

base materials are given in Table 4.5. After the wet etch, germanium,

gold, and nickel were evaporated onto the sample. The layers and the

thicknesses were Ge/Au/Ge/Au + Ni/Au (10.8/10.2/6.3/23.6 + 10/200


nm). Then the sample was soaked into the acetone. The photoresist

dissolves in the acetone and the metal on top of it was lifted-off. After

the cleaning, the sample was treated at 400 oC for 60 sec. Figure 4.7

shows the n+ ohmic contacts before and after the thermal process.

Table 4.5: Etch rates of some common etchants. The rates are indicated as nm/sec. Stop means no etch.

Etchant InGaAs InAlAs InP InGaAsP H3PO4 : H2O2 : H2O (1:3:40) 4.5 5.0 stop stop Citric acid : H2O2 (1:2) 2.1 stop stop H2SO4 : H2O2 : H2O (1:8:80) 8.3 stop H2SO4 : H2O2 : H2O (1:1:10) stop 1.7 HCl : H2O (3:1) stop 10.8 etches HCl : H2O (2:1) stop 133 HCl : H2O (1:1) stop 11.6 HCl : H3PO4 (1:1) stop 41.7 HCl : H3PO4 (1:3) stop 12.5

(a) (b) Figure 4.7: The n+ ohmic contacts before (a) and after (b) the RTA.

2. p+ Ohmic Contact Formation: After the cleaning, the photoresist was

exposed by using normal photolithography. Then Ti/Au (15/200 nm)

metals were evaporated onto the sample. The metals were lifted of

using the same method used in the previous step. After cleaning the


sample, it was thermally processed at 450 oC for 30 sec. We could not

increase the RTP temperature, because the quality of the n+ ohmic

contacts started to degrade. Figure 4.8 shows the p+ metal contacts

before deposition of the metal and after the thermal process.

(a) (b) Figure 4.8: The p+ ohmic contacts before the metal deposition (a) and after the

RTA.

(a) (b) Figure 4.9: (a) The color change on the surface is due to the variation of the etch depth (b) the mesa contour can be seen as a thin black line below the p

contact.

3. Mesa Isolation: With a normal lithography, the active area and the

ohmic contacts were covered with photoresist. Then the layers were

etched down to the undoped InAlAs layer for RCE wafer and down to


InP substrate for the single-pass wafer. The etch was done in several

steps and the etch depth was monitored with DEKTAK surface

profilometer. The sidewalls of the mesa had an inclination, which was

good for further metallization.

4. Surface Passivation: After the cleaning, the surface was covered with

Si3N4 layer. This layer was deposited at 250 oC with the PECVD

technique. Normal lithography was used to define the area to be etched

away. Dilute hydrofluoric acid solution was used to etch the dielectric

layer. The sidewalls and the active area of the photodetector was kept

covered, while the area that the interconnect metal will be deposited

was cleaned. Figure 4.10 shows the photograph of the devices before

and after the HF etches.

(a) (b) Figure 4.10: (a) The surface was covered with silicon nitride, and the

photograph shows the surface after photolithography, (b) the surface of the sample after the HF etch and the cleaning. Some area on top of the contacts was

open, and the sidewalls were covered.

5. Interconnect Metallization: Image-reversal photolithography was

used to define the metallization area. Then 0.7 ~ 0.8 µm thick Ti/Au

metals were evaporated onto the sample. The sample was soaked into


acetone, and the metals were lifted off from the unwanted area. The

Interconnect metals formed two connections to the ohmic contacts for

DC measurements. Also the ground-signal-ground type of the coplanar

waveguide was used during the high speed measurements. The metals

were on the highly resistive layers, and went over the insulating layer

deposited on the sidewalls of the mesa.

(a) (b)

(c) (d) Figure 4.11: (a, b) The photoresist before the metallization, (c, d) interconnect

metals after the cleaning.

After the interconnect metallization, the photodetectors were ready for the

measurements. Figure 4.12 shows SEM picture of these photodetectors.


(a)

(b)

Figure 4.12: SEM images of large area photodetectors.

4.2.2 Small Area Photodetector Fabrication The first three fabrication steps were same as the above fabrication. The rest of

the fabrication is as follows:

4. Interconnect Metalization: The photolithography and the metal

deposition were similar to the interconnect metallization used for large

area photodetector fabrication. Except the metals were deposited only

onto the highly resistive layer without connections to the ohmic

contacts.

5. Surface Passivation and Dielectric Deposition: After the cleaning,

the surface was covered with Si3N4 layer. This layer was deposited at

250 oC with the PECVD technique. Normal lithography was used to

define the area to be etched away. Dilute hydrofluoric acid solution was


used to etch the dielectric layer. The sidewalls and the active area of the

photodetector were kept covered. To make an electrical connection in

the final step, the top of the ohmic contacts and the nearest connection

point of the interconnect metals were cleaned from the dielectric. Also

the dielectric coating was etched from the area of the interconnect metal

where the high speed and DC probes are touched. This dielectric layer

was used for both passivation of the detector surface and formation of

the bias capacitors between the n+ ohmic contact of the photodetector

and the ground of the coplanar waveguide.

6. Post layer formation: After the sample cleaning, the surface was

covered with photoresist and normal lithography was used for

patterning. Then the sample was baked in the oven at 140 oC for 30

minutes to harden the photoresist. Then the photoresist was thinned

with RIE using oxygen gas.

7. Airbridge Formation: Without cleaning, a new layer of photoresist

was deposited on the sample. Using the image-reversal

photolithography, the connections were patterned on the sample. Then

0.7 ~ 0.8 µm thick Ti/Au metals were evaporated onto the sample. The

sample was soaked into acetone, and the metals were lifted off from the

unwanted area. With this metallization, the connections between the

ohmic contacts and the interconnect metals were formed, and also the

top plate of the biasing capacitor deposited. The schematics of this

fabrication step are shown in Figure 4.13.


hardenedphotoresist

hardened photoresist

substrate

ohmic metal

interconnect metal

Dielectric coating

n+ contact layer

p+ contact layerInGaAs active layer

photoresistphotoresistevaporated metal

evaporated metal evaporated metal

(a)

substrate

ohmic metal

interconnect metal

Dielectric coating

n+ contact layer

p+ contact layerInGaAs active layer

airbridge

(b)

Figure 4.13: Cross-section of the sample (a) after airbridge metallization, and (b) after the liftoff of the metal.

After the airbridge metallization, the photodetectors were ready for the

measurements. Figure 4.14 shows SEM picture of the airbridge connections,

while Figure 4.15 has the 3D illustration of the photodetector, and the

connections made during the current-voltage and responsivity measurements.

Figure 4.16 has detailed pictures of small area photodetector.


airbridge

detectormesa interconnect

metal

airbridge

detectormesa interconnect

metal

(a) (b) Figure 4.14: SEM picture of airbridges.

p+ InAlAs

InGaAs

n+ InAlAs

fiber

n+ ohmic

p+ ohmic

Interconnect metal Interconnect metal

Semi-Insulating InP Substrate

InAlAs/InAlGaAs Bottom DBR

DC probesDC probes

Figure 4.15: 3D illustration of a fabricated photodetector.

4.3 Measurements Properties of the photodetectors were measured at room (300 K) temperature.


(a)

(b)

(c)

(d)

Figure 4.16: Pictures of small area photodetectors obtained with optical microscope and SEM.


4.3.1 Current-Voltage Characteristics The current-voltage (I-V) measurements were done for different area

photodetectors between the reverse breakdown voltages to a few volts of

forward bias. The current limited not to damage the photodetectors. High

forward current caused damage by heating the contact between the interconnect

metal and the p+ ohmic metal. The reverse breakdown occurred around 14

volts of bias as shown on Figure 4.17 (a). This breakdown was mainly due to

the band-to-band tunneling. Figure 4.17 (b) shows the I-V measurement results

for different area photodetectors. The increase of the dark current with the area

can be clearly seen.

The analysis of dark current versus area showed that the dark current

density decreased linearly with the area for the detectors having diameter

greater than and equal to 60 µm. The dark current density of 30 µm diameter

photodetectors was 8×10-3 A/cm2 while 60 µm diameter photodetectors was

5×10-3 A/cm2. The dark current for smaller area detectors was similar to the 30

µm diameter photodetector. This showed us that the dark current was governed

by generation-recombination current in the surface depletion region. This

current can be decreased by using a better passivation layer, or can be

completely eliminated by using planar type photodetectors.

We also calculated the differential resistance (Rd) of the photodetectors

by simply differentiating the current-voltage data. The maximum value of Rd

was at 0.25 V reverse bias for all the photodetectors. The maximum value of Rd

had an exponential decay with the area, and the limit value for very large area

detectors was 188 MΩ. The differential resistance at zero bias (R0) had similar

behavior. The R0*A product was 310 Ω·cm2 for 200 µm diameter

photodetectors.


-14 -12 -10 -8 -6 -4 -2 0

-0.010

-0.008

-0.006

-0.004

-0.002

0.000

C

urre

nt (A

)

Voltage (V)

-1010-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

Cur

rent

(A)

Figure 4.17: (a) Cphotodetector, where t

logarithmic plot of diameter pho

(a)

)

(b

-8 -6 -4 -2 0

Bias Voltage (V)

urrent voltage characteristics of a 30 µm diameter he breakdown can be seen around -14 V bias. (b) The I-V characteristics of 30, 60, 100, 150, and 200 µm todetector (from bottom to top respectively)


0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4 2.5x10-4 3.0x10-4 3.5x10-40.0

2.0x10-9

4.0x10-9

6.0x10-9

8.0x10-9

1.0x10-8

1.2x10-8

1.4x10-8

Dar

k C

urre

nt @

-1 V

(A)

Area (cm2)

0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4 2.5x10-4 3.0x10-4 3.5x10-40.00

2.50x10-5

5.00x10-5

7.50x10-5

1.00x10-4

Dar

k C

urre

nt D

ensi

ty @

-1V

(A/c

m2 )

Area (cm2)

Figure 4.18: (a) Dark current of photodetectors at 1 V reverse bias as a function of detector area. (b) Dark current density of photodetectors at 1 V reverse bias

as a function of detector area. The linear decrease can be seen for detectors having 60 µm diameter.

(a)

(b)


-2.5 -2.0 -1.5 -1.0 -0.5 0.0105

106

107

108

109

Diff

eren

tial R

esis

tanc

e (Ω

)

Bias Voltage (V)

0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4 2.5x10-4 3.0x10-4 3.5x10-40.0

4.0x108

8.0x108

1.2x109

1.6x109

2.0x109Data: Data1_HModel: ExpDec1Equation: y = A1*exp(-x/t1) + y0Weighting: y Instrumental Chi^2/DoF = 0.70765R^2 = 0.94901 y0 188906257.65945 ±12606140.39048A1 1787413961.42312 ±345450527.40292t1 0.00006 ±7.2974E-6

Rd m

ax (Ω

)

Area (cm2)

Figure 4.19: (a) Calculated differential resistance of the photodetectors with 30, 60, 100, 150, and 200 µm diameter (from top to bottom respectively). (b) The

maximum differential resistance of photodetectors as a function of detector area.

(a)

(b)


0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4 2.5x10-4 3.0x10-4 3.5x10-40

1x107

2x107

3x107

4x107

R0 (Ω

)

Area (cm2)

0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4 2.5x10-4 3.0x10-4 3.5x10-4200

250

300

350

R0 *

A (Ω

cm

2 )

Area (cm2)

Figure 4.20: (a) Calculated differential resistance of the photodetectors at zero

bias. (b) R0A product as a function of detector area.

(a)

(b)


4.3.2 Responsivity Characteristics Photoresponse measurements were carried out in the 1530-1630 nm range

using a tunable laser source. The output of the laser was coupled to a single

mode fiber. The light was delivered to the devices by a lightwave fiber probe,

and the electrical characterization was carried out on a microwave probe

station. The top p+ layers were recess etched in small steps, and the tuning of

the resonance wavelength within the high reflectivity spectral region of the

DBR was observed.

Figure 4.21 (a) shows the spectral quantum efficiency measurements of

a device under 5 V reverse bias obtained by consecutive recess etches. Plot 1 is

the quantum efficiency after the top InGaAs layer etch, while plots 2, 3, 4, 5, 6,

and 7 correspond to cumulative recess etches of 80, 105, 150, 180, 210 and 240

nm, respectively. The peak experimental quantum efficiency 30% of the as-

grown sample at 1645 nm increases to 55% at 1614 nm after the first etch. The

peak quantum efficiency increased up to 66% with tuning until the resonance

wavelength reached 1572 nm. This increase was due to the increase of the

absorption coefficient of InGaAs at shorter wavelengths. As we continued the

recess etch, the peak quantum efficiency decreased due to the decrease of the

reflectivity of the Bragg mirror. The resonance wavelength was tuned for a

total of 47 nm (1538 – 1605 nm) while keeping the peak efficiencies above

60%. The peak efficiency was above 50% for the resonant wavelengths

between 1550 and 1620 nm, corresponding to a tuning range of 70 nm.

Figure 4.21 (b) shows the quantum efficiency measurement and

simulation results of the photodetector when the cavity resonance had been

tuned to 1572 nm. The difference between the simulation and the measurement

can be explained with the ±3% measurement error of the commercial

photodiode that was used for calibration purposes, and the uncertainty of the


layer thickness used during the simulation [72]. The full width at half

maximum (FWHM) of the devices was around 35 nm. The quantum efficiency

measurements were done at 5 V reverse bias under 0.5 mW optical input power.

When we increased the reverse bias beyond 3 V the active layer was fully

depleted, and the quantum efficiency increased 6% with respect to zero bias.

The responsivity of the PDs were also measured under various reverse

biases up to 7.4 mW optical power, which was the maximum power that could

be obtained from the laser. Figure 4.22 (a) shows the photocurrent versus input

optical power at the resonance wavelength of 1572 nm. Under 4 V and higher

reverse biases, the PDs had a linear photoresponse up to 7.4 mW optical power.

At 7.4 mW optical power, the device exhibited a 6.18 mA photocurrent. The

corresponding responsivity of the photodetectors was 0.835 A/W. The detector

responded linearly to the optical power in four orders of magnitude range under

4 V bias. Figure 4.22 (b) shows the photocurrent versus optical power for 0 and

4 V reverse bias values. The saturation was mainly due to the electric field

screening caused by photo-generated carriers inside the active layer [73].

4.3.3 High-Speed Characteristics High-speed measurements were made with a picosecond fiber laser operating

at 1550 nm. The 1 ps FWHM optical pulses from the laser were coupled to the

active area of the p-i-n photodiodes by means of a fiber probe. At zero bias, the

response of the photodetectors had a long tail due to the diffusion of the

carriers in the active layers. Measurements were done under bias to deplete the

active layer completely and to get rid of the diffusion tail. Above 5 V reverse

bias, we got a Gaussian response with a short tail. Figure 4.23 (a) shows the

temporal response of a small area (5×5 µm2) photodetector measured at 7 V

bias by a 50 GHz sampling scope.


1530 1550 1570 1590 1610 16300

10

20

30

40

50

60

70

7

6

54 3

21

Q

uant

um E

ffici

ency

(%)

Wavelength (nm)

1530 1550 1570 1590 1610 16300

10

20

30

40

50

60

70

Qua

ntum

Effi

cien

cy (%

)

Wavelength (nm)

Simulation Measurement

Figure 4.21: (a) The spectral quantum efficiency measurement for successive recess etches. (b) The measurement and calculation of the spectral quantum

efficiency with the peak at 1572 nm.

(a)

(b)


0 1 2 3 4 5 6 7 80

1

2

3

4

5

6

74 V

3 V

2 V

1 V

0 V

Pho

tocu

rren

t (m

A)

Optical Power (mW)

10-3 10-2 10-1 100 101

10-3

10-2

10-1

100

101

Pho

tocu

rren

t (m

A)

Optical Power (mW)

Figure 4.22: (a) The photocurrent vs optical power measurement of the photodetectors under various bias voltages. (b) Log-Log plot of the previous

data for 0 and 4 V bias voltages.

(a)

(b)


The photodiode output had 16 ps FWHM. The measured data was corrected by

deconvolving the effect of the 40 GHz bias-tee. After the deconvolution, the

device had a 3-dB bandwidth of 31 GHz. Larger area devices (80 µm2) also

showed similar responses, which showed that the temporal response was

limited by the transport of the photogenerated carriers. The 120 µm2 area

photodetectors had 28 GHz bandwidth. The measured bandwidth is lower than

the theoretically predicted 3-dB bandwidth of 45 GHz. Although grading layers

have been implemented to avoid carrier trapping, our measurement data shows

that the device performance is still limited by the carrier trapping. In our

devices, we used a digital grading that consisted of InP lattice matched

InGaAs/InAlAs layers. A linear grading may further improve the device

performance [74].

4.4 Conclusion

We have demonstrated high-speed, and high-efficiency resonant cavity

enhanced (RCE) InGaAs based p-i-n photodetectors. A peak quantum

efficiency of 66% was measured along with 28 GHz bandwidth, which

corresponds to 18.5 GHz bandwidth-efficiency product. The photoresponse

was linear up to 7.4 mW optical power, where the devices exhibited 6.18mA

photocurrent.


0 50 100 150 2000

5

10

15

20

25

30

35

40

Vol

tage

(mV

)

Time (psec)

FWHM = 16 ps

1 10

0.1

1

Freq

uenc

y R

espo

nse

Frequency (GHz)

3-dB frequency = 31 GHz

Figure 4.23: (a) The temporal response of 5×5 µm2 area photodetector under 7

V reverse bias. (b) The calculated frequency response of the photodetector.

(a)

(b)

Chapter 5

InSb p-i-n Photodetector

The spectral region above the visible range is called as the infrared spectrum.

The electromagnetic spectrum with the wavelength 1 µm and 14 µm is divided

into three regions that are separated by strong atmospheric absorption. These

absorptions are mainly due to carbon dioxide (CO2), carbon monoxide (CO),

and water (H2O) vapor present at the atmosphere. Near-infrared radiation

(near-IR) lies in the most energetic window, stretching from the upper edge of

the visible spectrum around 850 nm to a region of strong attenuation near 3 µm.

The mid-wavelength infrared (mid-IR) region lies between 3 and 5 µm, with

the long-wavelength infrared (long-IR) window stretching from the upper edge

of the broad absorption to an upper limit of 14 µm. Figure 5.1 shows the

spectral transmission of the atmosphere at the sea level.

Figure 5.1: Spectral transmission of the atmosphere at the sea level.

81

CHAPTER 5. INSB P-I-N PHOTODETECTOR 82

Infrared detectors can be used in thermal imaging systems, free space

communication, and chemical agent monitoring. The peak black body emission

from hot surfaces has the in the mid-IR spectrum. Hence military applications

include early warning systems, and night vision. Most of the carbon hydrates

(HC) such as methane, propane, and natural gases used for the heating, carbon

dioxide (CO2), carbon monoxide (CO), nitrous oxide (NOx) has absorption

band in the mid-IR region. Table 5.1 lists a few of these gases that have

absorption in the mid-IR region.

Table 5.1: Gases and their absorption properties.

Gas Absorption Wavelength (µm)

Methane (CH4) 3.20 ~ 3.45 Flammable Acetylene (C2H2) 2.99 ~ 3.04 Fire Ethylene (C2H4) 3.10 ~ 3.40 Fire Ethane (C2H6) 3.30 Fire Methyl Chloride (CH3Cl) 3.22 ~ 3.38 Fire Hydrogen Chloride (HCl) 3.33 ~ 3.70 Inhale Hazard Hydrogen Bromide (HBr) 3.82 Carbon Dioxide (CO2) 4.23 Carbon Monoxide (CO) 4.60 Nitric Oxide (NO) 5.25 Nitrous Oxide (N2O) 4.47 Nitrogen Dioxide (NO2) 6.14

The detection of these gases in the atmosphere has many industrial and

indoor applications, such as indoor air quality measurement, automotive and

flue gas emissions, hazardous are warning signals, gas leak detection, patient

monitoring, and alcohol breath analyzers. In all these applications, it is possible

to use non-dispersive infrared (NDIR) sensors. These sensors contain a light


source emitting light only in a specified spectrum, and an infrared detector to

measure the absorbance of the atmosphere. Another application is the free

space communication, which is becoming popular for short distance high speed

communication. These systems do not require cables or fibers, instead send the

optical signals directly in the atmosphere between two endpoints.

Although variable-bandgap HgCdTe has been used for many years in

military and commercial applications, it is being replaced by quantum well or

InSb photodetectors due to the technological difficulties [75]. Other possible

option is the quantum-well infrared photodetectors. But the normal incidence

operation is not possible for these detectors, unless special gratings are

integrated on the surface.

InSb is the common material for the mid-infrared (mid-IR) range

photodetectors with the cut-off wavelength of 5.4 µm at 77 K as shown in

Table 5.2. Ternary compounds of InSb has been used with the addition of As,

Bi, or Tl to extend the responsivity to the far infrared region [76,77]. Although

these detector structures can be fabricated on InSb, GaSb or InAs substrates,

this method is not preferred due to low resistance of the substrate and the

difficulty of thinning process for the back illuminated focal plane arrays [78-

80]. Instead Si and GaAs substrates are more attractive even there is a

significant lattice mismatch between the substrate and epitaxial layers [81-83].

Although optical and electrical properties have been studied over the years,

high-speed properties of InSb based photodetectors have not been reported

before in the scientific literature.

This chapter presents our effort to fabricate high speed and high

efficiency photodetector operating in the mid-infrared (MIR) wavelength range.


Table 5.2: Properties of some low energy gap semiconductors.

InSb InAs GaSb AlSb

Energy band-gap Eg (eV) 0.235 (0K) 0.170 (300K)

0.418 0.354

0.811 0.720

1.686 1.615

Cut-off wavelength λg (µm) 5.28 (0K) 7.29 (300K)

2.97 3.50

1.53 1.72

0.74 0.77

Band Type D D D I Crystal structure Z Z Z Z Lattice parameter (Å) 6.4794 6.0584 6.0959 6.1355 Density (g cm-3) 5.775 5.667 5.614 4.26 Melting point (K) 800 1215 985 1338 Intrinsic carrier conc. (cm-3) 2.0×1016 1.3×1015 Electron mobility (cm2 / V·s) 80000 33000 5000 200 Hole mobility (cm2 / V·s) 1250 460 850 420 Effective electron mass (m0) 0.0145 0.023 0.042 0.12 Effective hole mass (m0) 0.40 0.40 0.40 0.98

5.1 Design We designed a p-i-n type InSb based photodetector structure to be grown on

semi-insulating GaAs wafer. The detector structure was grown by molecular

beam epitaxy method by SVT Associates in USA. A 0.1 µm thick GaSb layer

was grown as a buffer between lattice mismatched GaAs and InSb layers. InSb

growth conditions and the thickness of the buffer layer were controlled using a

Reflection High Energy Electron Diffraction (RHEED). After the buffer layer a

1.5 µm thick n+ InSb layer, 1.5 µm thick n- InSb layer, and finally the 0.5 µm

thick p+ InSb layer were grown. The n- active layer was unintentionally doped

to 2×1015 cm-3. Tellurium and Beryllium were used as the n- and p- layer

dopants, respectively. The properties of these layers are also shown on Table

5.3. Doping level was 1018 cm-3 for both highly doped layers to decrease the

serial resistance. The thickness of the top p+ layer was kept lower than the n+

layer to increase the quantum efficiency.


Table 5.3: Epitaxial layers of InSb p-i-n photodetector.

Material Thickness (nm) Type Doping Concentration (cm-3) InSb 500 p+ Be 1×1018

InSb 1500 n- Te 2×1015

InSb 1500 n+ Te 1×1018

GaSb Buffer Layer Semi-Insulating GaAs Substrate

Figure 5.2: Lattice constant and energy gap of semiconductors.

5.2 Fabrication

Before a complete fabrication, we tried to optimize etch of InSb. Due to thick

epitaxial layers, deep etches had to be performed. RIE etch was impossible

with the gases present in our system due to nonselective etch of InSb and

photoresist. A few wet etch recipes were used for the optimization process.


• HNO3 : HCl : H2O (1:1:8) solution was use and no significant etch was

observed after 12 min etch.

• HNO3 : HF : H2O (1:1:6) solution was used. Reasonable etch rate was

observed at first, but this solution was not suitable for further etch due

to surface corrugation.

• H3PO4 : H2O2 : H2O solution had etch rates between 3 nm/sec (1:1:8) to

5 nm/sec (1:1:4). But this solution either non-uniformly etched InSb

and GaSb layers or did not etch some regions. Fig 5.3(a) shows the etch

profile of ohmic transmission line rectangle. The InSb layer was etched

more at the edges. Fig 5.3(b) shows the photograph of the same pattern.

All the layers can be seen as the result of non-uniform etch. Fig 5.3(c)

shows the area that has not been etched.

• Citric acid (C6H8O7): H2O2 (1:1) solution had the best result in terms of

uniformity. Citric acid was prepared mixing citric acid monohydrate

and water (100 g to 100 ml) at 80 oC. Then citric acid was cooled to

room temperature. During etch, the solution was kept at 30 oC, and

stirred for uniformity. The etch rate of the InSb was around 25 nm/min,

which is much lower than the above solutions. As the duration of etch

was high, the solution usually penetrates through the edge of the resist

and causes high under-etch. Thus the surface had to be protected by a

dielectric layer. A combination of Si3N4 and photoresist was used as the

mask during wet etches. This etch is isotropic and the under-etch and

the Si3N4 mask can be seen on the figure 5.3(d).


(a)

photoresist

InSb

GaSb

GaAs

photoresist

InSb

GaSb

GaAs

(b)

(c) (d) Figure 5.3: Details from InSb wet etch.

Before the fabrication, the surface was covered with 100 nm thick Si3N4 using

PECVD. This layer was kept until the end of the mesa etch step. The devices

were fabricated by a microwave-compatible process and completed in five

steps.

1) n+ ohmic contact formation: After the lithography, the Si3N4 layer

was etched with RIE. Using citric acid solution, the sample was etched

down to the top of the n+ InSb layer with a total depth of 2.4 µm. The

under-etch was around 2 µm. This amount of under-etch was not

critical for the consecutive contact. Titanium (12 nm) and gold (240 nm)

were deposited and lifted-off.


2) p+ ohmic contact formation: Following the lithography, Si3N4 was

etched using dilute hydro fluoric acid. After etch, titanium (12 nm) and

gold (160 nm) were deposited and lifted-off.

Semi-insulating GaAs substrate

GaSb buffer layer

n- InSb 1.5 µm

p+ InSb 0.5 µm

n+ InSb 1.5 µm

Photoresist

(a)


GaSb buffer layer

n+ InSb 1.5 µm

Photoresist

n- InSb 1.5 µm

p+ InSb 0.5 µm

(b)


GaSb buffer layer

n+ InSb 1.5 µm

Photoresist

n- InSb 1.5 µm

p+ InSb 0.5 µm

SiO2 isolation layer

(c)


GaSb buffer layer

n+ InSb 1.5 µm

n- InSb 1.5 µm

p+ InSb 0.5 µm


(d)

Figure 5.4: Modified mesa isolation step used during the fabrication of InSb photodetectors.

3) Mesa isolation etch: Active area and the n+ ohmic contacts were

protected with dielectric and photoresist. We used two different

methods for the mesa isolation. The first was similar to the fabrication

of the previous GaAs and InGaAs based photodetectors. All the layers

were etched down to the semi-insulating GaAs substrate. The total

height of the mesa was around 4 µm. The tall mesa was a major

problem for small area detectors, as it was impossible to make image-

reversal lithography, which was essential for thick interconnect


metallization. Due to this problem the mesa etches were modified. The

second method had a thinner etch and an additional dielectric

deposition. The layers were etched down to 2.4 µm. Without cleaning

the photoresist, 0.5 µm thick SiO2 layer was grown on the sample using

PECVD at 80 oC. This layer later acted as the insulating medium

between the highly conductive InSb layer and the interconnect metal.

Very thin dielectric layer was also deposited on top of the photoresist,

but the photoresist was lifted-off without problem. Finally the thin

Si3N4 layer was removed with diluted HF solution. These steps are

shown on Figure 5.4.

4) Dielectric deposition: The whole sample was covered with Si3N4 layer

that acts as both passivation and antireflection coating. After the

lithography, some regions on top of the ohmic contacts were etched

with diluted HF solution.

5) Interconnect metallization: After the image reversal lithography,

titanium (12 nm) and gold (800 nm) were deposited and lifted-off.

The photodetectors were fabricated with different active areas changing from

30 to 150 µm in diameter. Figure 5.5(a) shows the cross-section of a fabricated

photodetector, while the photograph of 150 µm diameter photodetector is

shown on figure 5.5(b).



GaSb buffer layer

n- InSb 1.5 µm

p+ InSb 0.5 µm

n+ InSb 1.5 µm


interconnectmetal

n+ ohmiccontact

silicon nitridecoating p+ ohmic

contact

interconnectmetal

(a)

interconnect metal to

p+ ohmic contactp+ ohmiccontact

n+ ohmiccontact

interconnect metal ton+ ohmic contactactive area

interconnect metal top+ ohmic contact

p+ ohmiccontact

n+ ohmiccontact

interconnect metal ton+ ohmic contactactive area

(b)

Figure 5.5: (a) Cross-section of a fabricated photodetector (b) Photograph of photodetector with 150 µm diameter.


5.3 Measurements

Properties of the photodetectors were measured both at room (300 K) and

liquid nitrogen (77 K) temperature.

5.3.1 Current-Voltage Characteristics Current-voltage (I-V) characteristics of the photodetectors were measured

using a Modular DC Source/Monitor Unit. Photodetectors were biased between

–2.0 and +0.5 volts at 300K background. Active area of the photodetectors

ranged from 7.06×10-6 cm2 (30 µm in diameter) to 2.25×10-4 cm2 (150×150

µm2). Initially I-V measurements were done at room temperature. Diode

characteristics were observed only at small area (30 and 60 µm diameter)

photodetectors. This is due to the high carrier concentration at the lightly doped

layer. Room temperature carrier concentration of InSb is above 2.0×1016 cm-3

for epitaxial layers grown on GaAs substrate due to lattice mismatch. This high

carrier concentration prevented us from observing a good p-n junction. Then

the temperature of the sample was decreased to 77 K.

Figure 5.6(a) shows the I-V characteristics of 30 µm diameter

photodetector. At room temperature dark current was 6.4 µA at zero bias and

4 mA at 1 V reverse bias. At 77 K, the dark current was 20 nA at zero bias and

41 µA at 1 V reverse bias. Figure 5.6 (b) shows the calculated differential

resistance of the photodetector. At room temperature differential resistance was

65 Ω at zero bias, and the highest value was 650 Ω at 440 mV reverse bias. The

peak position of the differential resistance shifted to higher bias voltages as the

temperature was lowered [84]. When the diode was cooled to 77 K, the

differential resistance at zero bias was 150 kΩ and the highest value was 170

kΩ at 40 mV forward bias.


-2.0 -1.5 -1.0 -0.5 0.0 0.510-8

10-7

10-6

10-5

10-4

10-3

10-2

Cur

rent

(A)

Voltage (V)

(a)

-2.0 -1.5 -1.0 -0.5 0.0 0.50

25

50

75

100

125

150

175

(b)

x100

Rd (k

Ω)

Voltage (V)

Figure 5.6: Current-voltage characteristics and differential resistance of 30 µm diameter photodetector at 300 K (dotted line) and 77 K (solid line).

-2.0 -1.5 -1.0 -0.5 0.0 0.510-8

10-7

10-6

10-5

10-4

10-3

10-2

60 µm 100 µm 150 µm

Cur

rent

(A)

Voltage (V)

Figure 5.7: Measured current-voltage characteristics of 60, 100, and 150 µm

diameter photodetectors at 77 K.

Figure 5.7 shows the I-V characteristics of cooled photodetectors with 60 µm,

100 µm, and 150 µm diameter. Dark current of the photodetectors gradually

increased with the increasing active area. Dark current at zero bias were 48 nA,


118 nA, and 239 nA for 60 µm, 100 µm, and 150 µm diameter photodetectors

respectively.

0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4 2.5x10-40

20

40

60

80

I d(µ

A)

Area (cm2)

0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4 2.5x10-40.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

J d(A/c

m2 )

Area (cm2)

Figure 5.8: (a) Dark current at -0.5 V as a function of area, while (b) shows the

dark current density (J0) as a function of area.

Figure 5.8(a) shows the dark current of the photodetectors at -0.5 V bias. A

linear increase can be seen for large area photodetectors. When the dark current

density was calculated at -0.5 V bias at 77 K, we saw that dark current density

was large for small area photodetectors. The calculated values can be seen on

Figure 5.8 (b). Dark current density (J0) analysis as a function of area showed

that the surface diffusion and generation current was dominant for low reverse


bias values [85]. The exponential fit to the J0 vs. area data showed that the

limit value for the large photodetectors was 0.34 A/cm2.

-2.0 -1.5 -1.0 -0.5 0.0 0.50

20

40

60

80

100

60 µm 100 µm 150 µm

Rd (k

Ω)

Voltage (V)

Figure 5.9: Calculated differential resistance (Rd) as a function of bias voltage.

Finally the differential resistance (Rd) of the photodetectors was

calculated by differentiating the current-voltage data. Figure 5.9 shows the Rd

as a function of bias voltage. The peak value of Rd was at 0 V bias, and

gradually decreased as we increased the reverse bias voltage. Zero bias

differential resistance (R0) showed an exponential decrease as a function of

area that is represented in Figure 5.10(a). According to this fit, R0 had almost

constant value of 20 kΩ for large area detectors. Zero-bias differential

resistance area product (R0A) for 30 µm diameter diodes was 1 Ω·cm2. This

product increased to 4.5 Ω·cm2 for 150×150 µm2 area diodes. Our values were

comparable to the reported R0A values. The R0A for InSb photodetectors

grown on Si were 1 Ω·cm2 (400×400 µm2 area) and 2.6 Ω·cm2 (400×80 µm2


area) [84,85]. R0A was 0.5 Ω·cm2 for 50×200 µm2 area InAsSb photodetector

grown on GaAs [82].

0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4 2.5x10-40

20

40

60

80

100

120

140

160

Ro (k

Ω)

Area (cm2)

0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4 2.5x10-40

1

2

3

4

5

R0A

(Ω c

m2 )

Area (cm2)

Figure 5.10: (a) The zero bias differential resistance (R0) as a function of

detector area. (b) Resistance-area product (R0A) as a function of detector area.

5.3.2 Responsivity Characteristics

Responsivity measurements were initially done at room temperature using a

tunable laser diode. The output of the laser was coupled to a single-mode fiber

and delivered to the active area of the photodetector using a fiber probe. The

photocurrent was recorded by digital ammeter while the detectors were biased

using a DC voltage source. First the responsivity of the photodetectors at 1550


nm was measured as a function of bias voltage. The responsivity increased

with the increase of the reverse bias as shown in Figure 5.11. Under 0.2 V

reverse bias the responsivity was 0.58 A/W, which corresponds to 46 %

quantum efficiency (QE). Due to the internal gain, the QE exceeded unity with

the increase of the reverse bias above 0.6 V. Spectral responsivity

measurements were made in the 1500 – 1600 nm spectral range which were

also shown at the inset of Figure 5.11. The photoresponse was nearly constant

in the given range.

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

1500 1525 1550 1575 16000.0

0.2

0.4

0.6

0.8

1.0

1.2

Res

pons

ivity

(A/W

)

Reverse Bias (V)

V = -0.6 V

V = -0.4 V

V = -0.2 V

Res

pons

ivity

(A/W

)

Wavelength (nm)

Figure 5.11: Responsivity of the photodetectors at 1550 nm as a function of the

reverse bias. The inset shows the spectral responsivity measurement under different reverse bias voltages ranging from 0.2 to 0.6 V.

These results were also in agreement with the spectral absorption

simulations based on the transfer matrix method. Figure 5.12 shows the

simulation results of reflection and absorption for the fabricated detector.

Below 1400 nm, most of the optical signal was absorbed in the upper p+ InSb


layer due to high absorption coefficient [86, 87]. Between 1400 and 2000 nm,

the absorption in the n- layer was nearly constant. Simulations show that a

further increase in the QE can be achieved by using a thinner p+ layer and by

tuning the anti-reflection coating to the desired wavelength.

800 1200 1600 2000 24000

10

20

30

40

50

60

70

0

10

20

30

40

50

60

70

Ref

lect

ion

(%)

Abso

rptio

n (%

)

Wavelength (nm)

n- Layer Absorption p+ Layer Absorption Reflection

Figure 5.12: Spectral simulation results for optical reflection and absorption in

the p+ and n- InSb layers.

After obtaining these results at room temperature, the measurements

were repeated at 77 K. Samples were mounted on an alumina substrate with

thermally conductive epoxy. Electrical connections were made with wire

bonding. Alumina holder was placed in a dewar with ZnSe optical window for

characterization. A blackbody at 774 K was used as the infrared (IR) light

source. The amplitude peak of the radiation from the source was at the mid-IR

wavelength region. Detector signal and noise were measured using a low noise

preamplifier and a lock-in amplifier, while the IR light was chopped at 680 Hz.


Figure 5.13: Picture of the setup from Electrical Engineering of METU.

The spectral measurements were made in 2.5 µm to 6.0 µm wavelength

range using an Oriel MIR-8000 FTIR system and a pyroelectric reference

detector. Figure 5.14 shows the 77 K spectral response of 80 µm (solid line)

and 60 µm (dotted line) diameter photodetectors. The cut-off wavelength

(where the responsivity dropped to half of the maximum) was 5.33 µm. This

corresponded to an energy bandgap of 0.23 eV, which was in good agreement

with the theoretical value at 77 K.

Responsivity decreased sharply after cut-off wavelength and it was two

orders of magnitude below the maximum value at 6.0 µm. In the responsivity

spectrum, two dips at 3.1 and 4.6 µm can be seen easily. This was due to the

absorption peaks of ice in the IR region [88, 89]. Due to leakage in the vacuum,

water vapor accumulated on the detector forming ice, which degraded the

performance of the detectors. Maximum responsivities of the detectors were


1.00×105 V/W and 4.41×104 V/W, respectively at 4.35 µm. Both

photodetectors had nearly equal current responsivity of 1.8 A/W and 1.7 A/W,

respectively which corresponds to 49 % quantum efficiency. We also measured

high detectivities for the photodetectors. Peak detectivities of the

photodetectors were 3.41×109 cm Hz1/2/W for 80 µm diameter detector and

7.98×109 cm Hz1/2/W for 60 µm diameter detector.

3 4 5 6

103

104

105

Res

pons

ivity

(V/W

)

Wavelength (µm)

Figure 5.14: Results of the spectral responsivity measurements are shown for 80 µm (solid line) diameter and 60 µm (dotted line) diameter photodetector.

5.3.3 High Speed Characteristics

Due to lack of suitable measurement setup, high-speed measurements had to be

done at room temperature. Picosecond full-width at half maximum (FWHM)

pulses were generated using a KTiOAsO4 (KTA) based optical parametric

oscillator (OPO) at 2500 nm. Figure 5.15 (a) shows the schematics of the OPO.


KTApump (780 nm)

signal (1133 nm)

filter

idler (2500 nm)

DUT

iris

KTApump (780 nm)

signal (1133 nm)

filter

idler (2500 nm)

DUT

iris

Figure 5.15: (a) The schematic diagram of the optical parametric oscillator, (b)

pictures of the experimental setup.

(a)

(b)

The OPO consisted of an optical resonator with four mirrors that were

highly reflective at the signal wavelength and a 20 mm long KTA crystal that


has been cut for non-critical phase matching along the θ = 90o and φ = 0o

direction. Type-II polarization geometry was employed to achieve parametric

creation [90]. OPO was pumped by a mode-locked Ti:Sapphire laser operating

at 780 nm wavelength with 150 femtosecond FWHM pulses at 76 MHz

repetition rate. Figure 5.15 (b) shows the pictures of the experimental setup.

Phase matching was achieved yielding a signal at 1133 nm and an idler at

around 2500 nm. At the output of the OPO the pump and the signal were

filtered out. The idler signal was then focused on the active area of the

photodetectors using an iris and infrared lens [68]. These pulses had a 1

picosecond FWHM.

The temporal response of the photodetectors was measured on a 50

GHz sampling scope and the detectors were biased using a 40 GHz bias-tee

[91]. Temporal responses of 30 and 60 µm diameter photodetectors were

measured as a function of reverse bias. Without bias, the responses had long

tails. This tail was due to the diffusion of carriers in the intrinsic region, which

could not be depleted without bias at room temperature. With the application of

bias voltage, we observed a reduction of the FWHM of the detector responses.

At 0.5 V bias, the FWHM values were measured as 59 and 104 psec for 30 and

60 µm diameter photodetectors respectively. After 1 V bias, FWHM values

decreased linearly with voltage up to 2.5 V. Measured FWHM values for the

detectors biased with 1.0 and 2.5 V were 41 and 33 psec for 30 µm, 65 and 40

psec for 60 µm. Figure 5.16 and figure 5.17 shows the temporal response of the

30 µm and 60 µm diameter photodetectors under 0.5 V, 1.0 V, and 2.5 V

biases. For each detector, both the FWHM and the fall time decreased as we

increased the bias voltage. Frequency responses of the detectors were

calculated using Fast Fourier Transform (FFT). Figure 5.18 (a) and (b) show


the calculated FFT results of the detectors as a function of bias voltage. Figure

5.18 (c) shows the 3-dB bandwidth of the detectors as a function of bias. The

linear increase in bandwidth with the bias can be seen easily beyond 1.0 V bias.

The maximum bandwidth measured for 30 µm diameter detector was 8.5 GHz.

Similar or better high-speed responses are expected when the photodetectors

are cooled to 77 K. The detector active area should be depleted easier and the

diffusion related slow responses can be eliminated.

Such high-speed infrared photodetectors can be used for optical

heterodyne detection and microwave mixing, infrared laser inspection, and free

space communication [92, 93]. It is known that mid-IR and far-IR wavelength

region (3-5 µm, and 8-14 µm) are better than visible or near infrared regions in

terms of transmission and background noise [94].

5.4 Conclusion

This study has shown the high speed operation of InSb based infrared detectors

for the first time in scientific literature [83, 95]. These detectors had 49 % peak

quantum efficiency at 4.35 µm. Also the detectivity was measured as 7.98×109

cm Hz1/2/W for 60 µm diameter detectors. Finally the 3-dB bandwidth of 30

µm diameter detectors was 8.5 GHz under 2.5 V reverse bias at room

temperature.


0 50 100 150 200 2500

10

20

30

40

tfall = 56 ps

33 ps

59 ps

41 ps

Time (psec)

0

10

20

30

40

tfall = 111 ps

Volta

ge (m

V)

0

10

20

30

40

tfall = 167 ps

(c)

(b)

(a)

Figure 5.16: Temporal response of a 30 µm diameter detector under (a) 0.5V,

(b) 1.0V, and (c) 2.5V bias.


0 100 200 300 400 5000

20

40

60

tfall = 81 ps

40 ps

104 ps

65 ps

Time (psec)

0

20

40

60

tfall = 168 ps

Volta

ge (m

V)

0

10

20

tfall = 900 ps

(c)

(b)

(a)

Figure 5.17: Temporal response of a 60 µm diameter detector under (a) 0.5V,

(b) 1.0V, and (c) 2.5V bias.


0.5 1.0 1.5 2.0 2.50

2

4

6

8

1 100.1

10.1

1

(c)

30 µm 60 µm

f 3-dB

(GH

z)

Bias (V)

(b)

0.5 V 1.0 V 2.5 V

Freq

uenc

y R

espo

nse

Frequency (GHz)

(a)

Figure 5.18: Fast Fourier Transform of the temporal responses of the

photodetectors. Results for (a) 60 µm diameter and (b) 30 µm diameter photodetectors as a function of bias are shown. (c) 3-dB bandwidth of the 30

and 60 µm diameter photodetectors as a function of applied bias.

Chapter 6

Conclusion and Suggestions for

Further Research

This thesis presented the design, fabrication, and characterization of high speed

and high efficiency infrared photodetectors. The design steps included detailed

material selection and optimization. The electrical properties also kept in mind

during the designs. After the growth, the wafers were characterized by optical

methods. Microwave compatible fabrication processes were made at Class-100

clean room environment at Bilkent University Department of Physics, while

measurement were done at the same laboratory in the Class-10000 clean room

environment

The first photodetector was based on InGaAs for the operation around

1550 nm wavelength. The resonant cavity enhancement (RCE) effect was

successfully applied to this detector design to improve the quantum efficiency.

Measurements showed that the detectors had 66% peak quantum efficiency at

1572 nm, which had 3 fold increases with respect to conventional detector

design. The detector had linear response up to 7.4 mW optical power under 4V

reverse bias. The high speed characteristics of the detectors showed 31 GHz

bandwidth for small area photodetectors. For 10x10 µm2 area detectors, those

had the full optical coupling from a single mode fiber; the 3-dB bandwidth was

106

CHAPTER 6. CONCLUSION AND SUGGESTIONS FOR FURTHER RE…107

measured as 28 GHz under 6V reverse bias. The corresponding bandwidth-

efficiency-product (BEP) was 18.5 GHz. Although this was the best BEP for

vertical RCE photodetectors, the characteristics of these detectors can be

further improved.

1. The p+ ohmic contact resistance was one of the limiting properties for

the high speed characteristics of the detectors. When this resistance is

lowered, higher bandwidths can be measured from these photodetectors.

A solution is to use different alloy for the contact, which would give

lower resistances when annealed at 450 oC.

2. The grading layers can be improved by using material grading between

InGaAs and InAlAs layers. This grading can be grown by using

different semiconductor layers having energy gap values sorted in

increasing order from InGaAs to InAlAs. This would overcome the

charge trapping problem at the interfaces.

The second detector was based on InSb. It was intended for operation

between 3 and 5 µm wavelengths. The detectors layers were grown on semi-

insulating GaAs substrates. Although there were a large lattice mismatch

between the substrate and the detector layers, very good electrical and optical

characteristics were observed at 77K. The detectors had low dark noise and

high differential resistance around zero bias. Also the responsivity

measurements showed 49% quantum efficiency. The detectivity was measured

as 7.98×109 cm Hz1/2/W for 60 µm diameter detectors. Finally the high speed

measurements showed 8.5 and 6.0 GHz bandwidth for 30 µm and 60 µm

diameter detectors, respectively. These are the first reported high speed

characteristics of InSb based photodetectors in the literature. To improve the

CHAPTER 6. CONCLUSION AND SUGGESTIONS FOR FURTHER RE…108

characteristics of these detectors, some modifications can be made during the

fabrication and measurements:

1. The high speed characteristics should be measured at 77 K to get the

true performance of the detectors. The bandwidth of the detectors must

increase at low temperatures, which corresponds to better

characteristics in terms of bandwidth.

2. The fabrication must be simplified to increase the fabrication yield. A

new mask should be designed without an interconnect metallization

along the sidewalls of the mesa. This metallization is hard made with

image reversal lithography due to large mesa heights.

In conclusion, the contribution of this research is the development of

design and fabrication methods to achieve high speed and highly efficient

vertical illuminated photodetectors.

BIBLIOGRAPHY

Bibliography [1] J. S. Kilby, “Turning potential into realities: The invention of the integrated

circuit,” From “Nobel Lectures, Physics 1996-2000,” Editor Gösta

Ekspong, (World Scientific Publishing Co., Singapore, 2002).

[2] Z. Alferov, “Double heterostructure lasers: early days and future

perspectives,” IEEE. J. Select Topics Quantum Electron., 6, 832 (2000).

[3]

D. Keck, “A future full of light,” IEEE. J. Select Topics Quantum

Electron., 6, 1254 (2000).

[4]

H. Kogelnik, “High-capacity optical communications: personal

recollections,” IEEE. J. Select Topics Quantum Electron., 6, 1279 (2000).

[5]

G. Cancellieri and F. Chiaraluce, “Recent Progress in Fibre Optics,” Prog.

Quant. Elect., 18, 39 (1994).

[6]

S. Y. Wang and D. M. Bloom, “100 GHz bandwidth planar GaAs Schottky

photodiode,” Electron. Lett., 19, 554 (1983).

[7]

E. Ozbay, K. D. Li, and D. M. Bloom, “2.0 psec GaAs monolithic

photodetector and all electronic sampler,” IEEE Photon. Tech. Lett., 3, 570

(1991).

[8]

K. D. Li, A. S. Hou, E. Ozbay, B. A. Auld, and D. M. Bloom, “2-

picosecond, GaAs photodiode optoelectronic circuit for optical correlation

applications,” Appl. Phys. Lett., 61, 3104 (1992).

109

BIBLIOGRAPHY

[9]

Y. G. Wey, K. S. Giboney, J. E. Bowers, M. J. Rodwell, P. Silvistre, P.

Thiagarajan, and G. Y. Ribbons, “110 GHz GaInAs/InP p-i-n Photodiodes

with Integrated Bias Tees and Matched Resistors,” IEEE Photon. Tech.

Lett., 5, 1310 (1993).

[10]

M. S. Unlu, S. Strite, “Resonant cavity enhanced photonic devices,” J.

Appl. Phys., 78, 607 (1995).

[11]

W. Schottky, “Halbleitertheorie der Sperrschicht,” Natursissenschaften, 26,

843 (1938).

[12]

B. L. Sharma, Metal-Semiconductor Schottky Barrier Junctions and Their

Applications (Plenum Press, 1984).

[13]

H. A. Bethe, “Theory of the Boundary Layer of Crystal Rectifiers,” MIT

Radiat. Lab. Rep., 43-12 (1942).

[14]

C. R. Crowell and S. M. Sze, “Current Transport in Metal-Semiconductor

Barriers,” Solid State Electron., 9, 1035 (1966).

[15] S. M. Sze, Physics of Semiconductor Devices (John Wiley & Sons, New

York, 1981).

[16]

E. Ozbay, M. S. Islam, B. Onat, M. Gokkavas, O. Aytur, G. Tuttle, E.

Towe, R. H. Herderson, M. S. Unlu, “Fabrication of High-Speed Resonant

Cavity Enhanced Schottky Photodiodes,” IEEE Photon. Technol. Lett., 9,

672 (1997).

[17]

B. M. Onat, M. Gokkavas, E. Ozbay, E. P. Ata, E. Towe, M. S. Unlu, “100-

GHz Resonant Cavity Enhanced Schottky Photodiodes,” IEEE Photon.

Technol. Lett., 10, 707 (1998).

110

BIBLIOGRAPHY

[18]

B. Corbett, L. Considine, S. Walsh and W. M. Kelly, “Narrow bandwith

long wavelength resonant cavity photodiodes,” Electron. Lett., 29, 2148

(1993).

[19]

A. Srinivasan, S. Murtaza, J. C. Campbell and B. G. Streetman, “High

quantum efficiency dual wavelength resonant-cavity photodetector,” Appl.

Phys. Lett., 66, 535 (1995).

[20]

E. Ozbay, I. Kimukin, N. Biyikli, O. Aytur, M. Gokkavas, G. Ulu, M. S.

Unlu, R. P. Mirin, K. A. Bertness, D. H. Christensen, “High-speed >90%

quantum-efficiency p-i-n photodiodes with a resonance wavelength

adjustable in the 795-835 nm range,” Appl. Phys. Lett., 74, 1072 (1999).

[21]

K. A. Anselm, S. S. Mustaza, C. Hu, H. Nie, B. G. Streetman, and J. C.

Campbell, “A Resonant-Cavity, Seperate-Apsorption-and-Multiplication,

Avalanche Photodiode with Low Excess Noise Factor,” IEEE Electron

Dev. Lett., 17, 91 (1997).

[22]

H. Nie, A. Anselm, C. Hu, S. S. Mustaza, B. G. Streetman, and J. C.

Campbell, “High-speed resonant-cavity seperate absorption and

multiplication avalanche photodiodes with 130 GHz gain-bandwith

product,” Appl. Phys. Lett., 70, 161 (1997).

[23] P. Yeh, Optical Waves in Layered Media (Wiley, New York, 1998).

[24]

L. A. Coldren and S. W. Corzine, Diode Lasers and Photonic Integrated

Circuits (Wiley, New York, 1995).

[25] M. Born and E. Wolf, Principals of Optics (Pergamon, 1980).

[26] G. Kinsley, C. Lenox, H. Nie, J. C. Campbell and B. G. Streetman,

111

BIBLIOGRAPHY

“Resonant Cavity Photodetector with Integrated Spectral Notch Filter,”

IEEE Photon. Technol. Lett., 10, 1142 (1998).

[27] M. D. Learmouth, N. P. Hewett, I. Reid, M. J. Robertson, “Integrated

Dielectric Wavelength Filters on InGaAs PIN Photodetectors,” Electron.

Lett., 26, 576 (1990).

[28]

T. Baba, S. Tamura, Y. Kokubin, and S. Watanabe, “Monolitic Integration

of Multilayer Filter on Vertival Surface of Semicionductor Substrate by a

Bias-Sputtering Technique,” IEEE Photon. Technol. Lett., 2, 191 (1990).

[29]

Datasheet “AZ 5214 E Image Reversal Photoresist,” by Clarient GmbH,

Germany available online at www.microchemicals.com.

[30] Imprint “Lithography process, and how to make them more reproducible,”

by Microchemicals GmbH, Germany available online at

www.microchemicals.com.

[31]

S. J. Pearton in InP HBTs: Growth, Processing, and Applications editors B.

Jalali and S. J. Pearton (Artech House, Boston, 1995).

[32] L. Zavieh, C. D. Nordquist, and T. S. Mayer, “Optimization of

In0.53Ga0.47As reactive ion etch with CH4/H2 using design of experiment

methods,” J. Vac. Sci. Technol. B., 16, 1024 (1998).

[33] R. Grover, J. V. Hryniewicz, O. S. King, and V. Van, “Process

development of methane-hydrogen-argon-based dry etching of InP for high

aspect-ratio structures with facet-quality sidewalls,” J. Vac. Sci. Technol.

B., 19, 1694 (2001).

[34] Y. B. Hahn, J. W. Lee, G. A. Vawter, R. J. Shul, C. R. Abernathy, D. C.

112

http://www.microchemicals.com/

http://www.microchemicals.com/

BIBLIOGRAPHY

Hays, E. S. Lambers, and S. J. Pearton, “Reactive ion beam etching of

GaAs ans related compounds in an inductive coupled plasma Cl2-Ar

mixture,” J. Vac. Sci. Technol. B., 17, 366 (1999).

[35] Necmi Biyikli, “High-performance AlxGa1-xN-based UV photodetectors for

visible/solar-blind applications,” Bilkent University PhD Thesis, May

2004.

[36]

Feridun Ay, “Silicon oxynitride layers for applications in optical

waveguides,” Bilkent University Ms Thesis, September 2000.

[37] Ralph Williams, Modern GaAs Processing Methods (Artech House, 1990).

[38]

Jin-Wei Shi, Kian-Giap Gan, Yi-Jen Chiu, Chi-Kuang Sun, Ying-Jay Yang,

and John E. Bowers, “Metal-semiconductor-metal traveling-wave

photodetectors,” IEEE Photon. Technol. Lett., 16, 623 (2001).

[39]

I-Hsing Tan, Chi-Kuang Sun, Kirk S. Giboney, John E. Bowers, Evelyn L.

Hu, B. I. Miller, and R. J. Capik, “120-GHz long-wavelength low-

capacitance photodetector with an air-bridge coplanar metal waveguide,”

IEEE Photon. Technol. Lett., 7, 1477 (1995).

[40]

E. Drödge, E. H. Böttcher, St. Kollakowski, A. Strittmatter, D. Bimberg, O.

Riemann, and R. Steingrüber, “78 GHz InGaAs MSM photodetector,”

Electron. Lett., vol. 34, no. 23, pp. 2241-2243, 1998.

[41]

G. S. Kinsey, C. C. Hansing, A. L. Holmes, Jr., B. G. Streetman, J. C.

Campbell, and A. G. Dentai, “Waveguide In0.53Ga0.47As-In0.52Al0.48As

Avalanche Photodiode,” IEEE Photon. Technol. Lett., 12, 416 (2000).

[42] S. Murthy, T. Jung, T. Chau, M. C. Wu, D. L. Sivco, and A. Y. Cho, “A

113

BIBLIOGRAPHY

novel monolithic distributed traveling-wave photodetector with parallel

optical feed,” IEEE Photon. Technol. Lett., 12, 681 (2000).

[43]

Y. Hirota, T. Hirono, T. Ishibashi, and H. Ito, “Traveling-wave

photodetector for 1.55 µm wavelength fabricated with unitraveling-carrier

photodiodes,” Appl. Phys. Lett., 78, 3767 (2001).

[44] K. Kato, A. Kozen, Y. Muramoto, Y. Itaya, T. Nagatsuma, and M. Yaita,

“110 GHz 50%-Efficiency Mushroom-Mesa Waveguide p-i-n Photodiode

for a 1.55-µm Wavelength,” IEEE Photon. Technol. Lett., 6, 719 (1994).

[45]

M. S. Unlu, K. Kishino, J. I. Chyi, L. Arsenault, J. Reed, and H. Morkoc,

“Wavelength demultiplexing heterojunction phototransistor,” Electron.

Lett., 26, 1857 (1990).

[46]

S. S. Murtaza, I. H. Tan, J. E. Bowers, E. H. Lu, K. A. Anselm, M. R.

Islam, R. V. Chelakara, R. D. Dupuis, B. G. Streetman, and J. C. Campbell,

“High-finesse resonant-cavity photodetectors with an adjustable resonance

frequency,” J. Lightwave Technol., 14, 1081 (1996).

[47]

H. Nie, K. A. Anselm, C. Hu, S. S. Murtaza, B. G. Streetman, and J. C.

Campbell, “High-speed resonant-cavity separate absorption and

multiplication avalanche photodiodes with 130 GHz gain-bandwidth

product,” Appl. Phys. Lett., 70, 161 (1997).

[48]

I. Kimukin, E. Ozbay, N. Biyikli, T. Kartaloğlu, O. Aytür, S. Ünlü, and G.

Tuttle, ”High-speed GaAs-based resonant-cavity-enhanced 1.3µm

photodetector,” Appl. Phys. Lett., 77, 3890 (2000).

[49] N. Biyikli, I. Kimukin, O. Aytur, M. Gokkavas, M. S. Unlu, E. Ozbay, “45-

114

BIBLIOGRAPHY

GHz bandwidth-efficiency resonant-efficiency-enhanced ITO-schottky

photodiodes,” IEEE Photon. Technol. Lett., 13, 705 (2001).

[50]

A. Strittmatter, S. Kollakowski, E. Drödge, E. H. Böttcher, and D.

Bimberg, “High speed, high efficiency resonant-cavity enhanced InGaAs

MSM photodetectors,” Electron. Lett., 32, 1231 (1996).

[51]

S. M. Spaziani, K. Vaccaro, and J. P. Lorenzo, “High-performance

substrate-removal InGaAs Schottky photodetectors,” IEEE Photon.

Technol. Lett., 10, 1144 (1998).

[52]

S. Y. Hu, J. Ko, and L. A. Coldren, “Resonant-cavity

InGaAs/InAlGaAs/InP photodetector arrays for wavelength demultiplexing

applications,” Appl. Phys. Lett., 70, 2347 (1997).

[53] W. A. Wohlmuth, J.-W Seo, P. Fay, C. Caneau, and I. Adesida “A high

speed ITO-InAlAs-InGaAs Schottky-barrier photodetector,” IEEE Photon.

Technol. Lett., 9. 1388 (1997).

[54]

C. Lennox, H. Nie, P. Yuan, G. Kinsey, A. L. Holmes, B. G. Streetman,

and J. C. Campbell, "Resonant-cavity InGaAs-InAlAs avalanche

photodiodes with gain-bandwidth product of 290 GHz," IEEE Photon.

Technol. Lett., 11, 1162 (1999).

[55] S. S. Murtaza, I-H Tan, R. V. Chelakara, M. R. Islam, A. Srinivasan, K. A.

Anselm, J. E. Bowers, E. L. Hu, R. D. Dupuis, B. G. Streetman, J. C.

Campbell, “High-efficiency, dual wavelength wafer-fused resoanat-cavity

photodetector operating at long wavelengths,” IEEE Photon. Technol. Lett.,

7, 679 (1995).

[56] Th. Engel, E. Drodge, G. Unterborsch, E. H. Bottcher, and D> Blimberg,

115

BIBLIOGRAPHY

“Reactive matching of millimeter-wave photodetectors using coplanar

waveguide technology,” Electron. Lett., 34, 1690 (1998).

[57] M. Horstmann, K. Schimpf, M. Marso, A. Fox, and P. Kordos, “16 GHz

bandwidth MSM photodetector and 45/85 fT/fmax HEMT prepared on an

identical InGaAs/InP layer structure,” Electron. Lett., 32, 763 (1996).

[58] J. H. Kim, H. T. Griem, R. A. Friedman, E. Y. Chan, ands. Ray, “High-

performance bak-illuminated InGaAs/InP MSM photodetector with a

record responsivity of 0.96 A/W,” IEEE Photon. Technol. Lett., 11, 1241

(1992).

[59] T. Chau, L. Fan, D. T. K. Tong, S. Mathai, M. C. Wu, D. L. Sivco, and A.

Y. Cho, “long wavelength velocity matched distributed photodetectors for

RF fibre optic links,” Electron. Lett., 34, 1422 (1998).

[60] M. S. Islam, S. Murthy, T. Itoh, M.C. Wu, D. Novak, R. Waterhouse, D.L.

Sivco, and A.Y. Cho, “Velocity-matched distributed Photodetectors and

Balanced Photodetectors with p-i-n Photodiodes,” IEEE Trans. Microwave

Theory Tech., 49, 1914 (2001).

[61] O. Madelung, Semiconductors – Basic Data (Springer, Berlin, 1996).

[62] D. A. Humphreys, R. J. King, D. Jenkins, and A. J. Moseley,

“Measurement of absorption coefficients of Ga0.47In0.53As over the

wavelength range 1.0-1.7 µm,” Electron. Lett., 21, 1187 (1985).

[63]

O. Blum, I. J. Fritz, L. R. Dawson, A. J. Howard, T. J. Headley, J. F. Klem,

and T. J. Drummond, “Highly reflective, long wavelength AlAsSb/GaAsSb

distributed Bragg reflector grown by molecular beam epitaxy in InP

substrates,” Appl. Phys. Lett., 66, 329 (1995).

116

BIBLIOGRAPHY

[64]

Sadao Adachi, “Refractive indices of III-V compounds: Key properties of

InGaAsP relevant to device design,” J. Appl. Phys., 53, 5863 (1982).

[65]

Sadao Adachi, “Optical dispersion relations for GaP, GaAs, GaSb, InP,

InAs, InSb, AlxGa1-xAs, and In1-xGaxAsyP1-y,” J. Appl. Phys., 66, 6030

(1989).

[66]

A. J. Moseley, J. Thompson, D. J. Robbins, and M. Q. Kearly, “High-

reflectivity AlGaInAs/InP multilayer mirrors grown by low-pressure

MOVPE for application to long-wavelength high-contrast-ratio multi-

quantum-well modulators,” Electron. Lett., 25, 1717 (1989).

[67]

J. –Ph. Debray, N. Bouché, G. Le Roux, R. Raj, and M. Quillec,

“Monolithic vertical cavity device lasing at 1.55µm in InGaAlAs system,”

Electron. Lett., 33, 868 (1997).

[68]

M. Guden, and J. Piprek, “Material parameters of quaternary III-V

semiconductors for multilayer mirrors and 1.55 µm wavelength,”

Modelling Simul. Mater. Sci. Eng., 4, 349 (1996).

[69]

Y. Takeda, in InP-Based Materials and Devices: Physics and Technology

edited by O. Wada, and H. Hasegawa (John Wiley & Sons, Inc., New York,

1999).

[70]

M. J. Mondry, D. I. Babic, J. E. Bowers, and L. A. Coldren, “Refractive

indexes of (Al, Ga, In)As epilayers on InP for optoelectronic applications,”

IEEE Photon. Tech. Lett., 4, 627 (1992).

[71]

C.C. Barron, C.J. Mahon, B.J. Thibeault, G. Wang, W. Jiang, L.A. Coldren,

and J.E. Bowers, “Resonant-cavity-enhanced pin photodetector with 17

117

BIBLIOGRAPHY

GHz bandwidth-efficiency product,” Electron. Lett., 30, 1796 (1994).

[72]

Newport Online Catalog. 818 Series Cylindrical detectors. [Online].

Available: http://www.newport.com.

[73]

L. Y. Lin, M. C. Wu, T. Itoh, T. A. Vang, R. E. Muller, D. L. Sivco, and A.

Y. Cho, “High-power high-speed photodetectors-Design, analysis, and

experimental demonstration,” IEEE Trans. Microwave Theory. Tech., 45,

1320 (1997).

[74] C. Lennox, H. Nie, G. Kinsey, P. Yuan, A. L. Holmes, Jr., B. G. Streetman,

and J. C. Campbell, “Improved optical response of superlattice graded

InAlAs/InGaAs p-i-n photodetectors,” Appl. Phys. Lett., 73, 3405 (1998).

[75]

A. Rogalski, “Quantum well photoconductors in infrared detector

technology,” J. Appl. Phys., 93, 4355 (2003).

[76]

J. J. Lee, J. D. Kim, and M. Razeghi, “Growth and characterization of

InSbBi for longwavelength infrared photodetectors,” Appl. Phys. Lett., 70,

3266 (1997).

[77]

J. D. Kim, S. Kim, D. Wu, J. Wojkowski, J. Xu, J. Piotrawski, E. Bigan,

and M. Razeghi, “8-13 µm InAsSb heterojunction photodiode operating at

near room temperature,” Appl. Phys. Lett., 67, 2645 (1995).

[78]

T. Ashley, A. B. Dean, C. T. Elliott, C. F. McConville, and C. R.

Whitehouse, “Molecular-beam growth of homoepitaxial InSb photovoltaic

detectors,” Electron. Lett., 24, 1270 (1988).

[79]

A. Rakovska, V. Berger, X. Marcadet, B. Vinter, G. Glestre, T.

Oksenhendler, and D. Kaplan, “Room temperature InAsSb photovoltaic

118

http://www.newport.com/

BIBLIOGRAPHY

midinfrared detector,” Appl. Phys. Lett., 77, 397 (2000).

[80]

D. T. Cheung, A. M. Andrews, E. R. Gertner, G. M. Williams, J. E. Clarke,

J. G. Pasko, and J. T. Longo, “Backside-illuminated InAs1-xSbx-InAs

narrow-band photodetectors,” Appl. Phys. Lett., 30, 587 (1977).

[81]

S. Ozer and C. Besikci, “Assessment of InSb photodetectors on Si

substarate,” J. Phys. D: Appl. Phys., 36, 559 (2003).

[82]

C. Besikci, S. Ozer, C. V. Hoof, L. Zimmermann, J. John, and P. Merken,

“Characteristics of InAs0.8Sb0.2 photodetectors on GaAs substarate,”

Semicond. Sci. Technol., 16, 992 (2001).

[83]

I. Kimukin, N. Biyikli, and E. Ozbay, “InSb high-speed photodetectors

grown on GaAs substrate,” J. Appl. Lett., 94, 5414 (2003).

[84]

E. Michel, J. Xu, J. D. Kim, I. Ferguson, and M. Razeghi, “InSb infrared

photodetectors on Si substrates grown by molecular beam epitaxy,” IEEE

Photon. Tech. Lett., 8, 673 (1996).

[85]

A. Tevke, C. Besikci, C. V. Hoof, and G. Borghs, “InSb infrared p-i-n

photodetectors grown on GaAs coated Si substrate by molecular beam

epitaxy,” Solid-State Electron., 42, 1039 (1998).

[86] Edward D. Palik, Handbook of Optical Constants of Solids (Academic

Press, Orlando, 1998), Vol. 1.

[87] T. S. Moss, in Semiconductors and Semimetals, edited by R. K. Willardson,

and A. C. Beer, (Academic Press, New York, 1966), Vol. 2.

[88]

S. G. Warren, “Optical constants of ice from ultraviolet to the microwave,”

Appl. Opt., 23, 1206 (1984).

119

BIBLIOGRAPHY

[89] David M. Wieliczka, Shengshan Weng, and Marvin R. Querry, “Wedge

shaped cell for highly absorbent liquids: infrared optical constants of

water,” Appl. Opt., 28, 1714 (1989).

[90] T. Kartaloğlu and O. Aytür, “Femtosecond self-doubling optical parametric

oscillator based on KTiOAsO4,” IEEE J. Quantum. Electron., 39, 65

(2003).

[91] I. Kimukin, N. Biyikli, B. Butun, O. Aytur, S. Ünlü, and E. Ozbay, ”

InGaAs Based High Performance p-i-n Photodiodes,” IEEE Photon. Tech.

Lett., 14, 366 (2002).

[92]

E. R. Brown, K. A. McIntosh, F. W. Smith, and M. J. Manfa, “Coherent

detection with GaAs/AlGaAs multiple quantum well structure,” Appl. Phys.

Lett., 62, 1513 (1993).

[93]

H. C. Liu, G. E. Jenkins, E. R. Brown, K. A. McIntosh, K. B. Nichols, and

M. J. Manfra, “Optical heterodyne detection and microwave rectification

up to 26 GHz using quantum well infrared photodetectors,” IEEE Electron.

Device Lett., 16, 253 (1995).

[94]

H. Manor and S. Arnon, “Performance of an optical wireless

communication as a function of wavelength,” Appl. Opt., 42, 4285 (2003).

[95]

I. Kimukin, N. Biyikli, T. Kartaloglu, O. Aytur, and E. Ozbay, “High-speed

InSb photodetectors on GaAs for Mid-IR applications,” IEEE. J. Select

Topics Quantum Electron., 10, 759 (2004).

120

Appendix A

Frequency Response calculation for a photodetector Based on the results by Barron et. al. (IEE Electron. Lett. vol.30 p.1796)

Clear@fD;sinc@x_D:= N@Sin@xDêxD;ε0= 8.85∗10^−14;ε= 12;µ= 1∗10^−4;

fRC@d_, A_, R_D:=1.73

2∗Pi∗R∗Hε∗ ε0∗AêdL ;

H∗ Transit time for electron and hole ∗Lte= 4.910^−12;th= 7.610^−12;

H∗ Capacitance limited frequency for the 50 ohm load ∗Lfrc= fRCA0.47∗µ, 200∗µ2, 50E;Print@" RC frequency = ", frcD;

H∗ find the 3−dB frequency of the photodetector ∗L

f3dB= FindRootA2==Hsinc@Pi∗f∗teD^2 + sinc@Pi∗f∗thD^2L^2

H1 + f^2 ê frc^2L ,

8f, 5∗10^9<E@@1, 2DD;

Print@"3−dB frequency = ", f3dBD;

RC frequency = 1.21854×1011

3−dB frequency = 4.55337×1010

121

Appendix B TMM based detector design This program is used during the design of the InGaAs based 1550nm pin RCE photodetector (* read the program file used for normal incidance *) <<"c:/ibrahim/mathematica/newoblique/normal.txt"; (* set the calculation parameters *) InitialWavelength=1400; FinalWavelength=1700; NumberOfWavelength=301; DataFileDirectory="c:/ibrahim/mathematica/material-data/"; Layers= "air", 0, DBR, 0, "si3n4", 223, "sio2", 267 , "ingaas", 40, "inalas", 310, "ingaas", 300, "inalas", 630, DBR, 25, "inalas", 121, "inalgaas", 112 , "inp", 0 ; (* calculate the electric field values *) InitializeStructure[]; (* Reflectivity calculation for the grown wafer *) refle=FindReflection[]; ListPlot[refle,PlotJoined→True,PlotStyle→RGBColor[0,0,1],Frame→True,FrameLabel→"Wavelength (nm)","Reflectivity (%)", PlotLabel->"Calculation for the reflectivity of InGaAs RCE pin wafer", TextStyle->FontFamily→"Arial",FontSize→12, ImageSize→550, GridLines→Automatic,Dashing[0.0075],Automatic,Dashing[0.0075]];

122

APPENDIX B 123

1400 1450 1500 1550 1600 1650 1700Wavelength H Lnm

0

20

40

60

80

ytivitcelfeR

H%L

Calculation for the reflectivity of InGaAs RCE pinwafer

Find the properties of the photodetector after the process is over and the tuning done (* make the simulation for the new thickness of the layers *) Layers= "air", 0, DBR, 0, "si3n4", 223, "sio2", 267 , "ingaas", 0, "inalas", 245, "ingaas", 300, "inalas", 630, DBR, 25, "inalas", 121, "inalgaas", 112 , "inp", 0 ; InitializeStructure[]; refle=FindReflection[]; ListPlot[refle,PlotJoined→True,PlotRange→-1,81,PlotStyle→RGBColor[0,0,1],Frame→True,FrameLabel→"Wavelength (nm)","Reflectivity (%)", PlotLabel->"Calculation for the reflectivity of InGaAs RCE pin Photodetector", TextStyle->FontFamily→"Arial",FontSize→12, ImageSize→550, GridLines→Automatic,Dashing[0.0075],Automatic,Dashing[0.0075]];

APPENDIX B 124

1400 1450 1500 1550 1600 1650 1700Wavelength HnmL

0

20

40

60

80

ytivitcelfeR

H%L

Calculation for the reflectivity of InGaAs RCE pinPhotodetector

abso=FindAbsorption[5]; ListPlot[abso,PlotJoined→True,PlotRange→-1,81,PlotStyle→RGBColor[0,0,1],Frame→True,FrameLabel→"Wavelength (nm)","Quantum Efficiency (%)", PlotLabel->"Calculation for the QE of InGaAs RCE pin Photodetector", TextStyle->FontFamily→"Arial",FontSize→12, ImageSize→550, GridLines→Automatic,Dashing[0.0075],Automatic,Dashing[0.0075]];

1400 1450 1500 1550 1600 1650 1700Wavelength HnmL

0

20

40

60

80

mutnauQ

ycneiciffEH%

L

Calculation for the QE of InGaAs RCEpin Photodetector

trans=FindTransmission[8];

APPENDIX B 125

ListPlot[trans,PlotJoined→True,PlotRange→-1,81, PlotStyle→RGBColor[0,0,1],Frame→True,FrameLabel→"Wavelength (nm)","Transmittivity (%)", PlotLabel->"Calculation for the transmittivity of InGaAs RCE pin Photodetector", TextStyle->FontFamily→"Arial",FontSize→12, ImageSize→550, GridLines→Automatic,Dashing[0.0075],Automatic,Dashing[0.0075]];

1400 1450 1500 1550 1600 1650 1700Wavelength HnmL

0

20

40

60

80

ytivittimsnarT

H%L

Calculation for the transmittivityof InGaAs RCE pinPhotodetector

Appendix C List of Publications

Our work on RCE photodetectors were mostly in the infrared region. This

work started with GaAs based photodetectors operating around 800 nm, and we

continued up to 1550 nm for different applications. The related publications

with increasing operating wavelength are as follows:

1. Ekmel Ozbay, Ibrahim Kimukin and Necmi Biyikli, "Ultrafast &

Highly Efficient Resonant Cavity Enhanced Photodiodes," Materials

Science Forum, volume 384-385, p. 241 (2002).

2. Ekmel Ozbay, Ibrahim Kimukin, Necmi Biyikli, Orhan Aytür, Mutlu

Gökkavas, Gökhan Ulu, R. Mirin, D. H. Christensen, and M.Selim

Ünlü, "High-Speed >90% Quantum-Efficiency p-i-n Photodiodes with

a Resonance Wavelength Adjustable in 795-835 nm Range," Applied

Physics Letters , volume 74, p. 1072 (1999).

3. Necmi Biyikli, Ibrahim Kimukin, Orhan Aytür, Mutlu Gokkavas, Selim

Ünlü, Ekmel Ozbay, "45-GHz Bandwidth-Efficiency Resonant-Cavity-

Enhanced ITO-Schottky Photodiodes," IEEE Photonics Technology

Letters , volume 13, p. 705 (2001).

4. Ibrahim Kimukin, Ekmel Ozbay, Necmi Biyikli, Tolga Kartaloglu,

Orhan Aytür, Selim Ünlü, Gary Tutle, "High-speed GaAs-based

126

APPENDIX C 127

resonant-cavity-enhanced 1.3µm photodetector," Applied Physics

Letters , volume 77, p. 3890 (2000).

5. Bayram Butun, Necmi Biyikli, Ibrahim Kimukin, Orhan Aytur, Ekmel

Ozbay, P. A. Postigo, J. P. Silveira, and A. R. Alija, "High-speed 1.55

µm operation of low-temperature-grown GaAs-based resonant-cavity-

enhanced p-i-n photodiodes," Applied Physice Letters, volume 84, p.

4185 (2004).

6. Ibrahim Kimukin, Necmi Biyikli, Bayram Butun, Orhan Aytur, Selim

Unlu, Ekmel Ozbay, "InGaAs Based High Performance p-i-n

Photodiodes," IEEE Photonics Technology Letters, volume 14, p. 366

(2002).

The publications for InSb based photodetectors are:

7. Ibrahim Kimukin, Necmi Biyikli, and Ekmel Ozbay, "InSb high-speed

photodetectors grown on GaAs substrate," Journal of Applied Physics,

volume 94, p. 5414 (2003).

8. Ibrahim Kimukin, Necmi Biyikli, Tolga Kartaloglu, Orhan Aytur, and

Ekmel Ozbay, "High-speed InSb photodetectors on GaAs for Mid-IR

applications," IEEE Journal of Selected Topics in Quantum Electronics,

volume 10, issue 4, p. 766 (2004).

In addition to infrared photodetectors, I also work on wide band-gap

semiconductors, especially GaN based photodetectors. My work was mostly on

the design, characterization, and initial process development for GaN based

structures. Our publications are as follows in chronological order:

9. Necmi Biyikli, Tolga Kartaloglu, Orhan Aytur, Ibrahim Kimukin,

Ekmel Ozbay, "High-speed visible-blind GaN-based indium–tin–oxide

APPENDIX C 128

Schottky photodiodes," Applied Physics Letters, volume 79, p. 2838

(2001).

10. Necmi Biyikli, Orhan Aytur, Ibrahim Kimukin, Turgut Tut, Ekmel

Ozbay, "Solar-blind AlGaN-based Schottky photodiodes with low noise

and high detectivity", Applied Physics Letters, volume 81, p. 3272

(2002).


Ekmel Ozbay, "High-Performance Solar-Blind AlGaN Schottky

Photodiodes," MRS Internet Journal of Nitride Semiconductor

Research, volume 8, p. 2 (2003).

12. Necmi Biyikli, Ibrahim Kimukin, Tolga Kartaloglu, Orhan Aytur,

Ekmel Ozbay, "High-speed Solar-Blind photodetectors with indium-tin-

oxide Schottky contacts," Applied Physics Letters, volume 82, p. 2344

(2003).


Ekmel Ozbay, "High-Speed Visible-Blind Resonant Cavity Enhanced

AlGaN Schottky Photodiodes," MRS Internet Journal of Nitride

Semiconductor Research, volume 8, p. 8 (2003).

14. Necmi Biyikli, Ibrahim Kimukin, Tolga Kartaloglu, Orhan Aytur,

Ekmel Ozbay, "High-Speed Colar-Blind AlGaN-based metal-

semiconductor-metal photodetectors," Physics Status Solidi C, volume

0, p. 2314 (2003).

15. Necmi Biyikli, Ibrahim Kimukin, Orhan Aytur, Ekmel Ozbay, "Solar-

blind AlGaN-based p-i-n photodiodes with low dark current and high

detectivity," IEEE Photonics Technology Letters, volume 16, p. 1718

(2004).

APPENDIX C 129

16. Necmi Biyikli, Ibrahim Kimukin, Turgut Tut, Tolga Kartaloglu, Orhan

Aytur, Ekmel Ozbay, "High-speed characterization of solar-blind

Al(x)Ga(1-x)N p-i-n photodiodes," Semiconductor Science and

Technology, volume 19, p. 1259 (2004).

17. Ekmel Ozbay, Necmi Biyikli, Ibrahim Kimukin, T. Tut, T. Kartaloglu,

and Orhan Aytur, "High-Performance Solar-Blind Photodetectors

Based on Al(x)Ga(1-x)N Heterostructures," IEEE Journal of Selected

Topics in Quantum Electronics, volume 10, issue 4, p. 742 (2004).

18. Necmi Biyikli, Ibrahim Kimukin, Bayram Butun, Orhan Aytur, and

Ekmel Ozbay, "ITO-Schottky Photodiodes for High-Performance

Detection in the UV-IR Spectrum," IEEE Journal of Selected Topics in

Quantum Electronics, volume 10, issue 4, p. 759 (2004).

19. Turgut Tut, Necmi Biyikli, Ibrahim Kimukin, Tolga Kartaloglu, Orhan

Aytur, M. Selim Unlu, and Ekmel Ozbay, "High bandwidth-efficiency

solar-blind AlGaN Schottky photodiodes with low dark current," Solid

State Electronics, volume 49, issue 1, p. 117 (2005).

HIGH SPEED AND HIGH EFFICIENCY INFRARED · PDF fileHIGH SPEED AND HIGH EFFICIENCY INFRARED...

Documents

Transcript of HIGH SPEED AND HIGH EFFICIENCY INFRARED · PDF fileHIGH SPEED AND HIGH EFFICIENCY INFRARED...